Merge branch 'main' into dataframe

Bump version to 0.0.1-a.20250805 in Cargo.toml

.github/htmldocs/index.html

@@ -58,6 +58,14 @@
         <h2>A lightweight dataframe & math toolkit for Rust</h2>
         <hr style="border: 1px solid #d4d4d4; margin: 20px 0;">
         <p>
+
+            🐙 <a href="https://github.com/Magnus167/rustframe">GitHub</a>
+            <br><br>
+            
+            📖 <a href="https://magnus167.github.io/rustframe/user-guide">User Guide</a>
+            <br><br>
+            
+
             📚 <a href="https://magnus167.github.io/rustframe/docs">Docs</a> |
             📊 <a href="https://magnus167.github.io/rustframe/benchmark-report/">Benchmarks</a>
 
@@ -65,8 +73,7 @@
             🦀 <a href="https://crates.io/crates/rustframe">Crates.io</a> |
             🔖 <a href="https://docs.rs/rustframe/latest/rustframe/">docs.rs</a>
             <br><br>
-            🐙 <a href="https://github.com/Magnus167/rustframe">GitHub</a> |
-            🌐 <a href="https://gitea.nulltech.uk/Magnus167/rustframe">Gitea mirror</a>
+            <!-- 🌐 <a href="https://gitea.nulltech.uk/Magnus167/rustframe">Gitea mirror</a> -->
         </p>
     </main>
 </body>

.github/workflows/docs-and-testcov.yml

@@ -153,7 +153,6 @@ jobs:
 
           echo "<meta http-equiv=\"refresh\" content=\"0; url=../docs/index.html\">" > target/doc/rustframe/index.html
 
-          mkdir output
           cp tarpaulin-report.html target/doc/docs/
           cp tarpaulin-report.json target/doc/docs/
           cp tarpaulin-badge.json target/doc/docs/
@@ -166,16 +165,30 @@ jobs:
           # copy the benchmark report to the output directory
           cp -r benchmark-report target/doc/
 
+          mkdir output
+          cp -r target/doc/* output/
+
+      - name: Build user guide
+        run: |
+          cargo binstall mdbook
+          bash ./docs/build.sh
+
+      - name: Copy user guide to output directory
+        run: |
+          mkdir output/user-guide
+          cp -r docs/book/* output/user-guide/
+
       - name: Add index.html to output directory
         run: |
-          cp .github/htmldocs/index.html target/doc/index.html
-          cp .github/rustframe_logo.png target/doc/rustframe_logo.png
+          cp .github/htmldocs/index.html output/index.html
+          cp .github/rustframe_logo.png output/rustframe_logo.png
 
       - name: Upload Pages artifact
         # if: github.event_name == 'push' || github.event_name == 'workflow_dispatch'
         uses: actions/upload-pages-artifact@v3
         with:
-          path: target/doc/
+          # path: target/doc/
+          path: output/
 
       - name: Deploy to GitHub Pages
         # if: github.event_name == 'push' || github.event_name == 'workflow_dispatch'

.github/workflows/run-unit-tests.yml

@@ -78,3 +78,8 @@ jobs:
         uses: codecov/test-results-action@v1
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
+
+      - name: Test build user guide
+        run: |
+          cargo binstall mdbook
+          bash ./docs/build.sh

.gitignore

@@ -17,3 +17,5 @@ data/
 tarpaulin-report.*
 
 .github/htmldocs/rustframe_logo.png
+
+docs/book/

Cargo.toml

@@ -1,11 +1,12 @@
 [package]
 name = "rustframe"
 authors = ["Palash Tyagi (https://github.com/Magnus167)"]
-version = "0.0.1-a.20250716"
+version = "0.0.1-a.20250805"
 edition = "2021"
 license = "GPL-3.0-or-later"
 readme = "README.md"
-description = "A simple dataframe library"
+description = "A simple dataframe and math toolkit"
+documentation = "https://magnus167.github.io/rustframe/"
 
 [lib]
 name = "rustframe"

README.md

@@ -1,11 +1,12 @@
 # rustframe
 
-📚 [Docs](https://magnus167.github.io/rustframe/) | 🐙 [GitHub](https://github.com/Magnus167/rustframe) | 🌐 [Gitea mirror](https://gitea.nulltech.uk/Magnus167/rustframe) | 🦀 [Crates.io](https://crates.io/crates/rustframe) | 🔖 [docs.rs](https://docs.rs/rustframe/latest/rustframe/)
+🐙 [GitHub](https://github.com/Magnus167/rustframe) | 📚 [Docs](https://magnus167.github.io/rustframe/) | 📖 [User Guide](https://magnus167.github.io/rustframe/user-guide/) | 🦀 [Crates.io](https://crates.io/crates/rustframe) | 🔖 [docs.rs](https://docs.rs/rustframe/latest/rustframe/)
 
 <!-- [![Last commit](https://img.shields.io/endpoint?url=https://magnus167.github.io/rustframe/rustframe/last-commit-date.json)](https://github.com/Magnus167/rustframe) -->
 
 [![codecov](https://codecov.io/gh/Magnus167/rustframe/graph/badge.svg?token=J7ULJEFTVI)](https://codecov.io/gh/Magnus167/rustframe)
 [![Coverage](https://img.shields.io/endpoint?url=https://magnus167.github.io/rustframe/docs/tarpaulin-badge.json)](https://magnus167.github.io/rustframe/docs/tarpaulin-report.html)
+[![gitea-mirror](https://img.shields.io/badge/git_mirror-blue)](https://gitea.nulltech.uk/Magnus167/rustframe)
 
 ---
 
@@ -152,7 +153,134 @@ let zipped_matrix = a.zip(&b, |x, y| x + y);
 assert_eq!(zipped_matrix.data(), &[6.0, 8.0, 10.0, 12.0]);
 ```
 
-### More examples
+---
+
+## DataFrame Usage Example
+
+```rust
+use chrono::NaiveDate;
+use rustframe::dataframe::DataFrame;
+use rustframe::utils::{BDateFreq, BDatesList};
+use std::any::TypeId;
+use std::collections::HashMap;
+
+// Helper for NaiveDate
+fn d(y: i32, m: u32, d: u32) -> NaiveDate {
+    NaiveDate::from_ymd_opt(y, m, d).unwrap()
+}
+
+// Create a new DataFrame
+let mut df = DataFrame::new();
+
+// Add columns of different types
+df.add_column("col_int1", vec![1, 2, 3, 4, 5]);
+df.add_column("col_float1", vec![1.1, 2.2, 3.3, 4.4, 5.5]);
+df.add_column(
+    "col_string",
+    vec![
+        "apple".to_string(),
+        "banana".to_string(),
+        "cherry".to_string(),
+        "date".to_string(),
+        "elderberry".to_string(),
+    ],
+);
+df.add_column("col_bool", vec![true, false, true, false, true]);
+// df.add_column("col_date", vec![d(2023,1,1), d(2023,1,2), d(2023,1,3), d(2023,1,4), d(2023,1,5)]);
+df.add_column(
+    "col_date",
+    BDatesList::from_n_periods("2023-01-01".to_string(), BDateFreq::Daily, 5)
+        .unwrap()
+        .list()
+        .unwrap(),
+);
+
+println!("DataFrame after initial column additions:\n{}", df);
+
+// Demonstrate frame re-use when adding columns of existing types
+let initial_frames_count = df.num_internal_frames();
+println!(
+    "\nInitial number of internal frames: {}",
+    initial_frames_count
+);
+
+df.add_column("col_int2", vec![6, 7, 8, 9, 10]);
+df.add_column("col_float2", vec![6.6, 7.7, 8.8, 9.9, 10.0]);
+
+let frames_after_reuse = df.num_internal_frames();
+println!(
+    "Number of internal frames after adding more columns of existing types: {}",
+    frames_after_reuse
+);
+assert_eq!(initial_frames_count, frames_after_reuse); // Should be equal, demonstrating re-use
+
+println!(
+    "\nDataFrame after adding more columns of existing types:\n{}",
+    df
+);
+
+// Get number of rows and columns
+println!("Rows: {}", df.rows()); // Output: Rows: 5
+println!("Columns: {}", df.cols()); // Output: Columns: 5
+
+// Get column names
+println!("Column names: {:?}", df.get_column_names());
+// Output: Column names: ["col_int", "col_float", "col_string", "col_bool", "col_date"]
+
+// Get a specific column by name and type
+let int_col = df.get_column::<i32>("col_int1").unwrap();
+// Output: Integer column: [1, 2, 3, 4, 5]
+println!("Integer column (col_int1): {:?}", int_col);
+
+let int_col2 = df.get_column::<i32>("col_int2").unwrap();
+// Output: Integer column: [6, 7, 8, 9, 10]
+println!("Integer column (col_int2): {:?}", int_col2);
+
+let float_col = df.get_column::<f64>("col_float1").unwrap();
+// Output: Float column: [1.1, 2.2, 3.3, 4.4, 5.5]
+println!("Float column (col_float1): {:?}", float_col);
+
+// Attempt to get a column with incorrect type (returns None)
+let wrong_type_col = df.get_column::<bool>("col_int1");
+// Output: Wrong type column: None
+println!("Wrong type column: {:?}", wrong_type_col);
+
+// Get a row by index
+let row_0 = df.get_row(0).unwrap();
+println!("Row 0: {:?}", row_0);
+// Output: Row 0: {"col_int1": "1", "col_float1": "1.1", "col_string": "apple", "col_bool": "true", "col_date": "2023-01-01", "col_int2": "6", "col_float2": "6.6"}
+
+let row_2 = df.get_row(2).unwrap();
+println!("Row 2: {:?}", row_2);
+// Output: Row 2: {"col_int1": "3", "col_float1": "3.3", "col_string": "cherry", "col_bool": "true", "col_date": "2023-01-03", "col_int2": "8", "col_float2": "8.8"}
+
+// Attempt to get an out-of-bounds row (returns None)
+let row_out_of_bounds = df.get_row(10);
+// Output: Row out of bounds: None
+println!("Row out of bounds: {:?}", row_out_of_bounds);
+
+// Drop a column
+df.drop_column("col_bool");
+println!("\nDataFrame after dropping 'col_bool':\n{}", df);
+
+println!("Columns after drop: {}", df.cols());
+println!("Column names after drop: {:?}", df.get_column_names());
+
+// Drop another column, ensuring the underlying Frame is removed if empty
+df.drop_column("col_float1");
+println!("\nDataFrame after dropping 'col_float1':\n{}", df);
+
+println!("Columns after second drop: {}", df.cols());
+println!(
+    "Column names after second drop: {:?}",
+    df.get_column_names()
+);
+
+// Attempt to drop a non-existent column (will panic)
+// df.drop_column("non_existent_col"); // Uncomment to see panic
+```
+
+## More examples
 
 See the [examples](./examples/) directory for some demonstrations of Rustframe's syntax and functionality.
 
@@ -191,10 +319,21 @@ cargo run --example
 
 Each demo runs a couple of mini-scenarios showcasing the APIs.
 
-### Running benchmarks
+## Running benchmarks
 
 To run the benchmarks, use:
 
 ```bash
 cargo bench --features "bench"
 ```
+
+## Building the user-guide
+
+To build the user guide, use:
+
+```bash
+cargo binstall mdbook
+bash docs/build.sh
+```
+
+This will generate the user guide in the `docs/book` directory.

docs/book.toml

@@ -0,0 +1,7 @@
+[book]
+title = "Rustframe User Guide"
+authors = ["Palash Tyagi (https://github.com/Magnus167)"]
+description = "Guided journey through Rustframe capabilities."
+
+[build]
+build-dir = "book"

docs/build.sh

@@ -0,0 +1,7 @@
+#!/usr/bin/env sh
+# Build and test the Rustframe user guide using mdBook.
+set -e
+
+cd docs
+bash gen.sh "$@"
+cd ..

docs/gen.sh

@@ -0,0 +1,14 @@
+#!/usr/bin/env sh
+
+set -e
+
+cargo clean
+
+cargo build --manifest-path ../Cargo.toml
+
+mdbook test -L ../target/debug/deps "$@"
+
+mdbook build "$@"
+
+cargo build
+# cargo build --release

docs/src/SUMMARY.md

@@ -0,0 +1,7 @@
+# Summary
+
+- [Introduction](./introduction.md)
+- [Data Manipulation](./data-manipulation.md)
+- [Compute Features](./compute.md)
+- [Machine Learning](./machine-learning.md)
+- [Utilities](./utilities.md)

docs/src/compute.md

@@ -0,0 +1,222 @@
+# Compute Features
+
+The `compute` module hosts numerical routines for exploratory data analysis.
+It covers descriptive statistics, correlations, probability distributions and
+some basic inferential tests.
+
+## Basic Statistics
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::stats::{mean, mean_horizontal, mean_vertical, stddev, median, population_variance, percentile};
+use rustframe::matrix::Matrix;
+
+let m = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+assert_eq!(mean(&m), 2.5);
+assert_eq!(stddev(&m), 1.118033988749895);
+assert_eq!(median(&m), 2.5);
+assert_eq!(population_variance(&m), 1.25);
+assert_eq!(percentile(&m, 50.0), 3.0);
+// column averages returned as 1 x n matrix
+let row_means = mean_horizontal(&m);
+assert_eq!(row_means.data(), &[2.0, 3.0]);
+let col_means = mean_vertical(&m);
+assert_eq!(col_means.data(), & [1.5, 3.5]);
+```
+
+### Axis-specific Operations
+
+Operations can be applied along specific axes (rows or columns):
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::stats::{mean_vertical, mean_horizontal, stddev_vertical, stddev_horizontal};
+use rustframe::matrix::Matrix;
+
+// 3x2 matrix
+let m = Matrix::from_rows_vec(vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0], 3, 2);
+
+// Mean along columns (vertical) - returns 1 x cols matrix
+let col_means = mean_vertical(&m);
+assert_eq!(col_means.shape(), (1, 2));
+assert_eq!(col_means.data(), &[3.0, 4.0]); // [(1+3+5)/3, (2+4+6)/3]
+
+// Mean along rows (horizontal) - returns rows x 1 matrix
+let row_means = mean_horizontal(&m);
+assert_eq!(row_means.shape(), (3, 1));
+assert_eq!(row_means.data(), &[1.5, 3.5, 5.5]); // [(1+2)/2, (3+4)/2, (5+6)/2]
+
+// Standard deviation along columns
+let col_stddev = stddev_vertical(&m);
+assert_eq!(col_stddev.shape(), (1, 2));
+
+// Standard deviation along rows
+let row_stddev = stddev_horizontal(&m);
+assert_eq!(row_stddev.shape(), (3, 1));
+```
+
+## Correlation
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::stats::{pearson, covariance};
+use rustframe::matrix::Matrix;
+
+let x = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+let y = Matrix::from_vec(vec![2.0, 4.0, 6.0, 8.0], 2, 2);
+let corr = pearson(&x, &y);
+let cov = covariance(&x, &y);
+assert!((corr - 1.0).abs() < 1e-8);
+assert!((cov - 2.5).abs() < 1e-8);
+```
+
+## Covariance
+
+### `covariance`
+
+Computes the population covariance between two equally sized matrices by flattening
+their values.
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::stats::covariance;
+use rustframe::matrix::Matrix;
+
+let x = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+let y = Matrix::from_vec(vec![2.0, 4.0, 6.0, 8.0], 2, 2);
+let cov = covariance(&x, &y);
+assert!((cov - 2.5).abs() < 1e-8);
+```
+
+### `covariance_vertical`
+
+Evaluates covariance between columns (i.e. across rows) and returns a matrix of
+column pair covariances.
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::stats::covariance_vertical;
+use rustframe::matrix::Matrix;
+
+let m = Matrix::from_rows_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+let cov = covariance_vertical(&m);
+assert_eq!(cov.shape(), (2, 2));
+assert!(cov.data().iter().all(|&v| (v - 1.0).abs() < 1e-8));
+```
+
+### `covariance_horizontal`
+
+Computes covariance between rows (i.e. across columns) returning a matrix that
+describes how each pair of rows varies together.
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::stats::covariance_horizontal;
+use rustframe::matrix::Matrix;
+
+let m = Matrix::from_rows_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+let cov = covariance_horizontal(&m);
+assert_eq!(cov.shape(), (2, 2));
+assert!(cov.data().iter().all(|&v| (v - 0.25).abs() < 1e-8));
+```
+
+### `covariance_matrix`
+
+Builds a covariance matrix either between columns (`Axis::Col`) or rows
+(`Axis::Row`). Each entry represents how two series co-vary.
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::stats::covariance_matrix;
+use rustframe::matrix::{Axis, Matrix};
+
+let data = Matrix::from_rows_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+
+// Covariance between columns
+let cov_cols = covariance_matrix(&data, Axis::Col);
+assert!((cov_cols.get(0, 0) - 2.0).abs() < 1e-8);
+
+// Covariance between rows
+let cov_rows = covariance_matrix(&data, Axis::Row);
+assert!((cov_rows.get(0, 1) + 0.5).abs() < 1e-8);
+```
+
+## Distributions
+
+Probability distribution helpers are available for common PDFs and CDFs.
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::stats::distributions::normal_pdf;
+use rustframe::matrix::Matrix;
+
+let x = Matrix::from_vec(vec![0.0, 1.0], 1, 2);
+let pdf = normal_pdf(x, 0.0, 1.0);
+assert_eq!(pdf.data().len(), 2);
+```
+
+### Additional Distributions
+
+Rustframe provides several other probability distributions:
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::stats::distributions::{normal_cdf, binomial_pmf, binomial_cdf, poisson_pmf};
+use rustframe::matrix::Matrix;
+
+// Normal distribution CDF
+let x = Matrix::from_vec(vec![0.0, 1.0], 1, 2);
+let cdf = normal_cdf(x, 0.0, 1.0);
+assert_eq!(cdf.data().len(), 2);
+
+// Binomial distribution PMF
+// Probability of k successes in n trials with probability p
+let k = Matrix::from_vec(vec![0_u64, 1, 2, 3], 1, 4);
+let pmf = binomial_pmf(3, k.clone(), 0.5);
+assert_eq!(pmf.data().len(), 4);
+
+// Binomial distribution CDF
+let cdf = binomial_cdf(3, k, 0.5);
+assert_eq!(cdf.data().len(), 4);
+
+// Poisson distribution PMF
+// Probability of k events with rate parameter lambda
+let k = Matrix::from_vec(vec![0_u64, 1, 2], 1, 3);
+let pmf = poisson_pmf(2.0, k);
+assert_eq!(pmf.data().len(), 3);
+```
+
+### Inferential Statistics
+
+Rustframe provides several inferential statistical tests:
+
+```rust
+# extern crate rustframe;
+use rustframe::matrix::Matrix;
+use rustframe::compute::stats::inferential::{t_test, chi2_test, anova};
+
+// Two-sample t-test
+let sample1 = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0, 5.0], 1, 5);
+let sample2 = Matrix::from_vec(vec![6.0, 7.0, 8.0, 9.0, 10.0], 1, 5);
+let (t_statistic, p_value) = t_test(&sample1, &sample2);
+assert!((t_statistic + 5.0).abs() < 1e-5);
+assert!(p_value > 0.0 && p_value < 1.0);
+
+// Chi-square test of independence
+let observed = Matrix::from_vec(vec![12.0, 5.0, 8.0, 10.0], 2, 2);
+let (chi2_statistic, p_value) = chi2_test(&observed);
+assert!(chi2_statistic > 0.0);
+assert!(p_value > 0.0 && p_value < 1.0);
+
+// One-way ANOVA
+let group1 = Matrix::from_vec(vec![1.0, 2.0, 3.0], 1, 3);
+let group2 = Matrix::from_vec(vec![2.0, 3.0, 4.0], 1, 3);
+let group3 = Matrix::from_vec(vec![3.0, 4.0, 5.0], 1, 3);
+let groups = vec![&group1, &group2, &group3];
+let (f_statistic, p_value) = anova(groups);
+assert!(f_statistic > 0.0);
+assert!(p_value > 0.0 && p_value < 1.0);
+```
+
+With the basics covered, explore predictive models in the
+[machine learning](./machine-learning.md) chapter.

docs/src/data-manipulation.md

@@ -0,0 +1,157 @@
+# Data Manipulation
+
+Rustframe's `Frame` type couples tabular data with
+column labels and a typed row index. Frames expose a familiar API for loading
+data, selecting rows or columns and performing aggregations.
+
+## Creating a Frame
+
+```rust
+# extern crate rustframe;
+use rustframe::frame::{Frame, RowIndex};
+use rustframe::matrix::Matrix;
+
+let data = Matrix::from_cols(vec![vec![1.0, 2.0], vec![3.0, 4.0]]);
+let frame = Frame::new(data, vec!["A", "B"], None);
+assert_eq!(frame["A"], vec![1.0, 2.0]);
+```
+
+## Indexing Rows
+
+Row labels can be integers, dates or a default range. Retrieving a row returns a
+view that lets you inspect values by column name or position.
+
+```rust
+# extern crate rustframe;
+# extern crate chrono;
+use chrono::NaiveDate;
+use rustframe::frame::{Frame, RowIndex};
+use rustframe::matrix::Matrix;
+
+let d = |y, m, d| NaiveDate::from_ymd_opt(y, m, d).unwrap();
+let data = Matrix::from_cols(vec![vec![1.0, 2.0], vec![3.0, 4.0]]);
+let index = RowIndex::Date(vec![d(2024, 1, 1), d(2024, 1, 2)]);
+let mut frame = Frame::new(data, vec!["A", "B"], Some(index));
+assert_eq!(frame.get_row_date(d(2024, 1, 2))["B"], 4.0);
+
+// mutate by row key
+frame.get_row_date_mut(d(2024, 1, 1)).set_by_index(0, 9.0);
+assert_eq!(frame.get_row_date(d(2024, 1, 1))["A"], 9.0);
+```
+
+## Column operations
+
+Columns can be inserted, renamed, removed or reordered in place.
+
+```rust
+# extern crate rustframe;
+use rustframe::frame::{Frame, RowIndex};
+use rustframe::matrix::Matrix;
+
+let data = Matrix::from_cols(vec![vec![1, 2], vec![3, 4]]);
+let mut frame = Frame::new(data, vec!["X", "Y"], Some(RowIndex::Range(0..2)));
+
+frame.add_column("Z", vec![5, 6]);
+frame.rename("Y", "W");
+let removed = frame.delete_column("X");
+assert_eq!(removed, vec![1, 2]);
+frame.sort_columns();
+assert_eq!(frame.columns(), &["W", "Z"]);
+```
+
+## Aggregations
+
+Any numeric aggregation available on `Matrix` is forwarded to `Frame`.
+
+```rust
+# extern crate rustframe;
+use rustframe::frame::Frame;
+use rustframe::matrix::{Matrix, SeriesOps};
+
+let frame = Frame::new(Matrix::from_cols(vec![vec![1.0, 2.0], vec![3.0, 4.0]]), vec!["A", "B"], None);
+assert_eq!(frame.sum_vertical(), vec![3.0, 7.0]);
+assert_eq!(frame.sum_horizontal(), vec![4.0, 6.0]);
+```
+
+## Matrix Operations
+
+```rust
+# extern crate rustframe;
+use rustframe::matrix::Matrix;
+
+let data1 = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+let data2 = Matrix::from_vec(vec![5.0, 6.0, 7.0, 8.0], 2, 2);
+
+let sum = data1.clone() + data2.clone();
+assert_eq!(sum.data(), vec![6.0, 8.0, 10.0, 12.0]);
+
+let product = data1.clone() * data2.clone();
+assert_eq!(product.data(), vec![5.0, 12.0, 21.0, 32.0]);
+
+let scalar_product = data1.clone() * 2.0;
+assert_eq!(scalar_product.data(), vec![2.0, 4.0, 6.0, 8.0]);
+
+let equals = data1 == data1.clone();
+assert_eq!(equals, true);
+```
+
+### Advanced Matrix Operations
+
+Matrices support a variety of advanced operations:
+
+```rust
+# extern crate rustframe;
+use rustframe::matrix::{Matrix, SeriesOps};
+
+// Matrix multiplication (dot product)
+let a = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+let b = Matrix::from_vec(vec![5.0, 6.0, 7.0, 8.0], 2, 2);
+let product = a.matrix_mul(&b);
+assert_eq!(product.data(), vec![23.0, 34.0, 31.0, 46.0]);
+
+// Transpose
+let m = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+let transposed = m.transpose();
+assert_eq!(transposed.data(), vec![1.0, 3.0, 2.0, 4.0]);
+
+// Map function over all elements
+let m = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+let squared = m.map(|x| x * x);
+assert_eq!(squared.data(), vec![1.0, 4.0, 9.0, 16.0]);
+
+// Zip two matrices with a function
+let a = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+let b = Matrix::from_vec(vec![5.0, 6.0, 7.0, 8.0], 2, 2);
+let zipped = a.zip(&b, |x, y| x + y);
+assert_eq!(zipped.data(), vec![6.0, 8.0, 10.0, 12.0]);
+```
+
+### Matrix Reductions
+
+Matrices support various reduction operations:
+
+```rust
+# extern crate rustframe;
+use rustframe::matrix::{Matrix, SeriesOps};
+
+let m = Matrix::from_rows_vec(vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0], 3, 2);
+
+// Sum along columns (vertical)
+let col_sums = m.sum_vertical();
+assert_eq!(col_sums, vec![9.0, 12.0]); // [1+3+5, 2+4+6]
+
+// Sum along rows (horizontal)
+let row_sums = m.sum_horizontal();
+assert_eq!(row_sums, vec![3.0, 7.0, 11.0]); // [1+2, 3+4, 5+6]
+
+// Cumulative sum along columns
+let col_cumsum = m.cumsum_vertical();
+assert_eq!(col_cumsum.data(), vec![1.0, 4.0, 9.0, 2.0, 6.0, 12.0]);
+
+// Cumulative sum along rows
+let row_cumsum = m.cumsum_horizontal();
+assert_eq!(row_cumsum.data(), vec![1.0, 3.0, 5.0, 3.0, 7.0, 11.0]);
+```
+
+With the basics covered, continue to the [compute features](./compute.md)
+chapter for statistics and analytics.

docs/src/introduction.md

@@ -0,0 +1,40 @@
+# Introduction
+
+🐙 [GitHub](https://github.com/Magnus167/rustframe) | 📚 [Docs](https://magnus167.github.io/rustframe/) | 📖 [User Guide](https://magnus167.github.io/rustframe/user-guide/) | 🦀 [Crates.io](https://crates.io/crates/rustframe) | 🔖 [docs.rs](https://docs.rs/rustframe/latest/rustframe/)
+
+Welcome to the **Rustframe User Guide**. Rustframe is a lightweight dataframe
+and math toolkit for Rust written in 100% safe Rust. It focuses on keeping the
+API approachable while offering handy features for small analytical or
+educational projects.
+
+Rustframe bundles:
+
+- column‑labelled frames built on a fast column‑major matrix
+- familiar element‑wise math and aggregation routines
+- a growing `compute` module for statistics and machine learning
+- utilities for dates and random numbers
+
+```rust
+# extern crate rustframe;
+use rustframe::{frame::Frame, matrix::{Matrix, SeriesOps}};
+
+let data = Matrix::from_cols(vec![vec![1.0, 2.0], vec![3.0, 4.0]]);
+let frame = Frame::new(data, vec!["A", "B"], None);
+
+// Perform column wise aggregation
+assert_eq!(frame.sum_vertical(), vec![3.0, 7.0]);
+```
+
+## Resources
+
+- [GitHub repository](https://github.com/Magnus167/rustframe)
+- [Crates.io](https://crates.io/crates/rustframe) & [API docs](https://docs.rs/rustframe)
+- [Code coverage](https://codecov.io/gh/Magnus167/rustframe)
+
+This guide walks through the main building blocks of the library. Each chapter
+contains runnable snippets so you can follow along:
+
+1. [Data manipulation](./data-manipulation.md) for loading and transforming data
+2. [Compute features](./compute.md) for statistics and analytics
+3. [Machine learning](./machine-learning.md) for predictive models
+4. [Utilities](./utilities.md) for supporting helpers and upcoming modules

docs/src/machine-learning.md

@@ -0,0 +1,282 @@
+# Machine Learning
+
+The `compute::models` module bundles several learning algorithms that operate on
+`Matrix` structures. These examples highlight the basic training and prediction
+APIs. For more end‑to‑end walkthroughs see the examples directory in the
+repository.
+
+Currently implemented models include:
+
+- Linear and logistic regression
+- K‑means clustering
+- Principal component analysis (PCA)
+- Gaussian Naive Bayes
+- Dense neural networks
+
+## Linear Regression
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::models::linreg::LinReg;
+use rustframe::matrix::Matrix;
+
+let x = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 4, 1);
+let y = Matrix::from_vec(vec![2.0, 3.0, 4.0, 5.0], 4, 1);
+let mut model = LinReg::new(1);
+model.fit(&x, &y, 0.01, 100);
+let preds = model.predict(&x);
+assert_eq!(preds.rows(), 4);
+```
+
+## K-means Walkthrough
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::models::k_means::KMeans;
+use rustframe::matrix::Matrix;
+
+let data = Matrix::from_vec(vec![1.0, 1.0, 5.0, 5.0], 2, 2);
+let (model, _labels) = KMeans::fit(&data, 2, 10, 1e-4);
+let new_point = Matrix::from_vec(vec![0.0, 0.0], 1, 2);
+let cluster = model.predict(&new_point)[0];
+```
+
+## Logistic Regression
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::models::logreg::LogReg;
+use rustframe::matrix::Matrix;
+
+let x = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 4, 1);
+let y = Matrix::from_vec(vec![0.0, 0.0, 1.0, 1.0], 4, 1);
+let mut model = LogReg::new(1);
+model.fit(&x, &y, 0.1, 200);
+let preds = model.predict_proba(&x);
+assert_eq!(preds.rows(), 4);
+```
+
+## Principal Component Analysis
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::models::pca::PCA;
+use rustframe::matrix::Matrix;
+
+let data = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+let pca = PCA::fit(&data, 1, 0);
+let transformed = pca.transform(&data);
+assert_eq!(transformed.cols(), 1);
+```
+
+## Gaussian Naive Bayes
+
+Gaussian Naive Bayes classifier for continuous features:
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::models::gaussian_nb::GaussianNB;
+use rustframe::matrix::Matrix;
+
+// Training data with 2 features
+let x = Matrix::from_rows_vec(vec![
+    1.0, 2.0,
+    2.0, 3.0,
+    3.0, 4.0,
+    4.0, 5.0
+], 4, 2);
+
+// Class labels (0 or 1)
+let y = Matrix::from_vec(vec![0.0, 0.0, 1.0, 1.0], 4, 1);
+
+// Train the model
+let mut model = GaussianNB::new(1e-9, true);
+model.fit(&x, &y);
+
+// Make predictions
+let predictions = model.predict(&x);
+assert_eq!(predictions.rows(), 4);
+```
+
+## Dense Neural Networks
+
+Simple fully connected neural network:
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::models::dense_nn::{DenseNN, DenseNNConfig, ActivationKind, InitializerKind, LossKind};
+use rustframe::matrix::Matrix;
+
+// Training data with 2 features
+let x = Matrix::from_rows_vec(vec![
+    0.0, 0.0,
+    0.0, 1.0,
+    1.0, 0.0,
+    1.0, 1.0
+], 4, 2);
+
+// XOR target outputs
+let y = Matrix::from_vec(vec![0.0, 1.0, 1.0, 0.0], 4, 1);
+
+// Create a neural network with 2 hidden layers
+let config = DenseNNConfig {
+    input_size: 2,
+    hidden_layers: vec![4, 4],
+    output_size: 1,
+    activations: vec![ActivationKind::Sigmoid, ActivationKind::Sigmoid, ActivationKind::Sigmoid],
+    initializer: InitializerKind::Uniform(0.5),
+    loss: LossKind::MSE,
+    learning_rate: 0.1,
+    epochs: 1000,
+};
+let mut model = DenseNN::new(config);
+
+// Train the model
+model.train(&x, &y);
+
+// Make predictions
+let predictions = model.predict(&x);
+assert_eq!(predictions.rows(), 4);
+```
+
+## Real-world Examples
+
+### Housing Price Prediction
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::models::linreg::LinReg;
+use rustframe::matrix::Matrix;
+
+// Features: square feet and bedrooms
+let features = Matrix::from_rows_vec(vec![
+    2100.0, 3.0,
+    1600.0, 2.0,
+    2400.0, 4.0,
+    1400.0, 2.0,
+], 4, 2);
+
+// Sale prices
+let target = Matrix::from_vec(vec![400_000.0, 330_000.0, 369_000.0, 232_000.0], 4, 1);
+
+let mut model = LinReg::new(2);
+model.fit(&features, &target, 1e-8, 10_000);
+
+// Predict price of a new home
+let new_home = Matrix::from_vec(vec![2000.0, 3.0], 1, 2);
+let predicted_price = model.predict(&new_home);
+println!("Predicted price: ${}", predicted_price.data()[0]);
+```
+
+### Spam Detection
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::models::logreg::LogReg;
+use rustframe::matrix::Matrix;
+
+// 20 e-mails × 5 features = 100 numbers (row-major, spam first)
+let x = Matrix::from_rows_vec(
+    vec![
+        // ─────────── spam examples ───────────
+        2.0, 1.0, 1.0, 1.0, 1.0, // "You win a FREE offer - click for money-back bonus!"
+        1.0, 0.0, 1.0, 1.0, 0.0, // "FREE offer! Click now!"
+        0.0, 2.0, 0.0, 1.0, 1.0, // "Win win win - money inside, click…"
+        1.0, 1.0, 0.0, 0.0, 1.0, // "Limited offer to win easy money…"
+        1.0, 0.0, 1.0, 0.0, 1.0, // ...
+        0.0, 1.0, 1.0, 1.0, 0.0, // ...
+        2.0, 0.0, 0.0, 1.0, 1.0, // ...
+        0.0, 1.0, 1.0, 0.0, 1.0, // ...
+        1.0, 1.0, 1.0, 1.0, 0.0, // ...
+        1.0, 0.0, 0.0, 1.0, 1.0, // ...
+        // ─────────── ham examples ───────────
+        0.0, 0.0, 0.0, 0.0, 0.0, // "See you at the meeting tomorrow."
+        0.0, 0.0, 0.0, 1.0, 0.0, // "Here's the Zoom click-link."
+        0.0, 0.0, 0.0, 0.0, 1.0, // "Expense report: money attached."
+        0.0, 0.0, 0.0, 1.0, 1.0, // ...
+        0.0, 1.0, 0.0, 0.0, 0.0, // "Did we win the bid?"
+        0.0, 0.0, 0.0, 0.0, 0.0, // ...
+        0.0, 0.0, 0.0, 1.0, 0.0, // ...
+        1.0, 0.0, 0.0, 0.0, 0.0, // "Special offer for staff lunch."
+        0.0, 0.0, 0.0, 0.0, 0.0, // ...
+        0.0, 0.0, 0.0, 1.0, 0.0,
+    ],
+    20,
+    5,
+);
+
+// Labels: 1 = spam, 0 = ham
+let y = Matrix::from_vec(
+    vec![
+        1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, // 10 spam
+        0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, // 10 ham
+    ],
+    20,
+    1,
+);
+
+// Train
+let mut model = LogReg::new(5);
+model.fit(&x, &y, 0.01, 5000);
+
+// Predict
+// e.g. "free money offer"
+let email_data = vec![1.0, 0.0, 1.0, 0.0, 1.0];
+let email = Matrix::from_vec(email_data, 1, 5);
+let prob_spam = model.predict_proba(&email);
+println!("Probability of spam: {:.4}", prob_spam.data()[0]);
+```
+
+### Iris Flower Classification
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::models::gaussian_nb::GaussianNB;
+use rustframe::matrix::Matrix;
+
+// Features: sepal length and petal length
+let x = Matrix::from_rows_vec(vec![
+    5.1, 1.4, // setosa
+    4.9, 1.4, // setosa
+    6.2, 4.5, // versicolor
+    5.9, 5.1, // virginica
+], 4, 2);
+
+let y = Matrix::from_vec(vec![0.0, 0.0, 1.0, 2.0], 4, 1);
+let names = vec!["setosa", "versicolor", "virginica"];
+
+let mut model = GaussianNB::new(1e-9, true);
+model.fit(&x, &y);
+
+let sample = Matrix::from_vec(vec![5.0, 1.5], 1, 2);
+let predicted_class = model.predict(&sample);
+let class_name = names[predicted_class.data()[0] as usize];
+println!("Predicted class: {} ({:?})", class_name, predicted_class.data()[0]);
+```
+
+### Customer Segmentation
+
+```rust
+# extern crate rustframe;
+use rustframe::compute::models::k_means::KMeans;
+use rustframe::matrix::Matrix;
+
+// Each row: [age, annual_income]
+let customers = Matrix::from_rows_vec(
+    vec![
+        25.0, 40_000.0, 34.0, 52_000.0, 58.0, 95_000.0, 45.0, 70_000.0,
+    ],
+    4,
+    2,
+);
+
+let (model, labels) = KMeans::fit(&customers, 2, 20, 1e-4);
+
+let new_customer = Matrix::from_vec(vec![30.0, 50_000.0], 1, 2);
+let cluster = model.predict(&new_customer)[0];
+println!("New customer belongs to cluster: {}", cluster);
+println!("Cluster labels: {:?}", labels);
+```
+
+For helper functions and upcoming modules, visit the
+[utilities](./utilities.md) section.

docs/src/utilities.md

@@ -0,0 +1,63 @@
+# Utilities
+
+Utilities provide handy helpers around the core library. Existing tools
+include:
+
+- Date utilities for generating calendar sequences and business‑day sets
+- Random number generators for simulations and testing
+
+## Date Helpers
+
+```rust
+# extern crate rustframe;
+use rustframe::utils::dateutils::{BDatesList, BDateFreq, DatesList, DateFreq};
+
+// Calendar sequence
+let list = DatesList::new("2024-01-01".into(), "2024-01-03".into(), DateFreq::Daily);
+assert_eq!(list.count().unwrap(), 3);
+
+// Business days starting from 2024‑01‑02
+let bdates = BDatesList::from_n_periods("2024-01-02".into(), BDateFreq::Daily, 3).unwrap();
+assert_eq!(bdates.list().unwrap().len(), 3);
+```
+
+## Random Numbers
+
+The `random` module offers deterministic and cryptographically secure RNGs.
+
+```rust
+# extern crate rustframe;
+use rustframe::random::{Prng, Rng};
+
+let mut rng = Prng::new(42);
+let v1 = rng.next_u64();
+let v2 = rng.next_u64();
+assert_ne!(v1, v2);
+```
+
+## Stats Functions
+
+```rust
+# extern crate rustframe;
+use rustframe::matrix::Matrix;
+use rustframe::compute::stats::descriptive::{mean, median, stddev};
+
+let data = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0, 5.0], 1, 5);
+
+let mean_value = mean(&data);
+assert_eq!(mean_value, 3.0);
+
+let median_value = median(&data);
+assert_eq!(median_value, 3.0);
+
+let std_value = stddev(&data);
+assert_eq!(std_value, 2.0_f64.sqrt());
+```
+
+Upcoming utilities will cover:
+
+- Data import/export helpers
+- Visualization adapters
+- Streaming data interfaces
+
+Contributions to these sections are welcome!

src/compute/mod.rs

@@ -1,3 +1,16 @@
+//! Algorithms and statistical utilities built on top of the core matrices.
+//!
+//! This module groups together machine‑learning models and statistical helper
+//! functions. For quick access to basic statistics see [`stats`](crate::compute::stats), while
+//! [`models`](crate::compute::models) contains small learning algorithms.
+//!
+//! ```
+//! use rustframe::compute::stats;
+//! use rustframe::matrix::Matrix;
+//!
+//! let m = Matrix::from_vec(vec![1.0, 2.0, 3.0], 3, 1);
+//! assert_eq!(stats::mean(&m), 2.0);
+//! ```
 pub mod models;
 
 pub mod stats;

src/compute/models/activations.rs

@@ -1,3 +1,15 @@
+//! Common activation functions used in neural networks.
+//!
+//! Functions operate element-wise on [`Matrix`] values.
+//!
+//! ```
+//! use rustframe::compute::models::activations::sigmoid;
+//! use rustframe::matrix::Matrix;
+//!
+//! let x = Matrix::from_vec(vec![0.0], 1, 1);
+//! let y = sigmoid(&x);
+//! assert!((y.get(0,0) - 0.5).abs() < 1e-6);
+//! ```
 use crate::matrix::{Matrix, SeriesOps};
 
 pub fn sigmoid(x: &Matrix<f64>) -> Matrix<f64> {

src/compute/models/dense_nn.rs

@@ -1,3 +1,30 @@
+//! A minimal dense neural network implementation for educational purposes.
+//!
+//! Layers operate on [`Matrix`] values and support ReLU and Sigmoid
+//! activations. This is not meant to be a performant deep‑learning framework
+//! but rather a small example of how the surrounding matrix utilities can be
+//! composed.
+//!
+//! ```
+//! use rustframe::compute::models::dense_nn::{ActivationKind, DenseNN, DenseNNConfig, InitializerKind, LossKind};
+//! use rustframe::matrix::Matrix;
+//!
+//! // Tiny network with one input and one output neuron.
+//! let config = DenseNNConfig {
+//!     input_size: 1,
+//!     hidden_layers: vec![],
+//!     output_size: 1,
+//!     activations: vec![ActivationKind::Relu],
+//!     initializer: InitializerKind::Uniform(0.5),
+//!     loss: LossKind::MSE,
+//!     learning_rate: 0.1,
+//!     epochs: 1,
+//! };
+//! let mut nn = DenseNN::new(config);
+//! let x = Matrix::from_vec(vec![1.0, 2.0], 2, 1);
+//! let y = Matrix::from_vec(vec![2.0, 3.0], 2, 1);
+//! nn.train(&x, &y);
+//! ```
 use crate::compute::models::activations::{drelu, relu, sigmoid};
 use crate::matrix::{Matrix, SeriesOps};
 use crate::random::prelude::*;

src/compute/models/gaussian_nb.rs

@@ -1,3 +1,16 @@
+//! Gaussian Naive Bayes classifier for dense matrices.
+//!
+//! ```
+//! use rustframe::compute::models::gaussian_nb::GaussianNB;
+//! use rustframe::matrix::Matrix;
+//!
+//! let x = Matrix::from_vec(vec![1.0, 2.0, 1.0, 2.0], 2, 2); // two samples
+//! let y = Matrix::from_vec(vec![0.0, 1.0], 2, 1);
+//! let mut model = GaussianNB::new(1e-9, false);
+//! model.fit(&x, &y);
+//! let preds = model.predict(&x);
+//! assert_eq!(preds.rows(), 2);
+//! ```
 use crate::matrix::Matrix;
 use std::collections::HashMap;
 

src/compute/models/k_means.rs

@@ -1,3 +1,14 @@
+//! Simple k-means clustering working on [`Matrix`] data.
+//!
+//! ```
+//! use rustframe::compute::models::k_means::KMeans;
+//! use rustframe::matrix::Matrix;
+//!
+//! let data = Matrix::from_vec(vec![1.0, 1.0, 5.0, 5.0], 2, 2);
+//! let (model, labels) = KMeans::fit(&data, 2, 10, 1e-4);
+//! assert_eq!(model.centroids.rows(), 2);
+//! assert_eq!(labels.len(), 2);
+//! ```
 use crate::compute::stats::mean_vertical;
 use crate::matrix::Matrix;
 use crate::random::prelude::*;

src/compute/models/linreg.rs

@@ -1,3 +1,16 @@
+//! Ordinary least squares linear regression.
+//!
+//! ```
+//! use rustframe::compute::models::linreg::LinReg;
+//! use rustframe::matrix::Matrix;
+//!
+//! let x = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 4, 1);
+//! let y = Matrix::from_vec(vec![2.0, 3.0, 4.0, 5.0], 4, 1);
+//! let mut model = LinReg::new(1);
+//! model.fit(&x, &y, 0.01, 100);
+//! let preds = model.predict(&x);
+//! assert_eq!(preds.rows(), 4);
+//! ```
 use crate::matrix::{Matrix, SeriesOps};
 
 pub struct LinReg {

src/compute/models/logreg.rs

@@ -1,3 +1,16 @@
+//! Binary logistic regression classifier.
+//!
+//! ```
+//! use rustframe::compute::models::logreg::LogReg;
+//! use rustframe::matrix::Matrix;
+//!
+//! let x = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 4, 1);
+//! let y = Matrix::from_vec(vec![0.0, 0.0, 1.0, 1.0], 4, 1);
+//! let mut model = LogReg::new(1);
+//! model.fit(&x, &y, 0.1, 100);
+//! let preds = model.predict(&x);
+//! assert_eq!(preds[(0,0)], 0.0);
+//! ```
 use crate::compute::models::activations::sigmoid;
 use crate::matrix::{Matrix, SeriesOps};
 

src/compute/models/mod.rs

@@ -1,3 +1,19 @@
+//! Lightweight machine‑learning models built on matrices.
+//!
+//! Models are intentionally minimal and operate on the [`Matrix`](crate::matrix::Matrix) type for
+//! inputs and parameters.
+//!
+//! ```
+//! use rustframe::compute::models::linreg::LinReg;
+//! use rustframe::matrix::Matrix;
+//!
+//! let x = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 4, 1);
+//! let y = Matrix::from_vec(vec![2.0, 3.0, 4.0, 5.0], 4, 1);
+//! let mut model = LinReg::new(1);
+//! model.fit(&x, &y, 0.01, 1000);
+//! let preds = model.predict(&x);
+//! assert_eq!(preds.rows(), 4);
+//! ```
 pub mod activations;
 pub mod dense_nn;
 pub mod gaussian_nb;

src/compute/models/pca.rs

@@ -1,3 +1,14 @@
+//! Principal Component Analysis using covariance matrices.
+//!
+//! ```
+//! use rustframe::compute::models::pca::PCA;
+//! use rustframe::matrix::Matrix;
+//!
+//! let data = Matrix::from_rows_vec(vec![1.0, 1.0, 2.0, 2.0], 2, 2);
+//! let pca = PCA::fit(&data, 1, 0);
+//! let projected = pca.transform(&data);
+//! assert_eq!(projected.cols(), 1);
+//! ```
 use crate::compute::stats::correlation::covariance_matrix;
 use crate::compute::stats::descriptive::mean_vertical;
 use crate::matrix::{Axis, Matrix, SeriesOps};

src/compute/stats/correlation.rs

@@ -1,3 +1,16 @@
+//! Covariance and correlation helpers.
+//!
+//! This module provides routines for measuring the relationship between
+//! columns or rows of matrices.
+//!
+//! ```
+//! use rustframe::compute::stats::correlation;
+//! use rustframe::matrix::Matrix;
+//!
+//! let x = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+//! let cov = correlation::covariance(&x, &x);
+//! assert!((cov - 1.25).abs() < 1e-8);
+//! ```
 use crate::compute::stats::{mean, mean_horizontal, mean_vertical, stddev};
 use crate::matrix::{Axis, Matrix, SeriesOps};
 

src/compute/stats/descriptive.rs

@@ -1,3 +1,15 @@
+//! Descriptive statistics for matrices.
+//!
+//! Provides means, variances, medians and other aggregations computed either
+//! across the whole matrix or along a specific axis.
+//!
+//! ```
+//! use rustframe::compute::stats::descriptive;
+//! use rustframe::matrix::Matrix;
+//!
+//! let m = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+//! assert_eq!(descriptive::mean(&m), 2.5);
+//! ```
 use crate::matrix::{Axis, Matrix, SeriesOps};
 
 pub fn mean(x: &Matrix<f64>) -> f64 {

src/compute/stats/distributions.rs

@@ -1,3 +1,16 @@
+//! Probability distribution functions applied element-wise to matrices.
+//!
+//! Includes approximations for the normal, uniform and gamma distributions as
+//! well as the error function.
+//!
+//! ```
+//! use rustframe::compute::stats::distributions;
+//! use rustframe::matrix::Matrix;
+//!
+//! let x = Matrix::from_vec(vec![0.0], 1, 1);
+//! let pdf = distributions::normal_pdf(x.clone(), 0.0, 1.0);
+//! assert!((pdf.get(0,0) - 0.3989).abs() < 1e-3);
+//! ```
 use crate::matrix::{Matrix, SeriesOps};
 
 use std::f64::consts::PI;

src/compute/stats/inferential.rs

@@ -1,3 +1,14 @@
+//! Basic inferential statistics such as t‑tests and chi‑square tests.
+//!
+//! ```
+//! use rustframe::compute::stats::inferential;
+//! use rustframe::matrix::Matrix;
+//!
+//! let a = Matrix::from_vec(vec![1.0, 2.0], 2, 1);
+//! let b = Matrix::from_vec(vec![1.1, 1.9], 2, 1);
+//! let (t, _p) = inferential::t_test(&a, &b);
+//! assert!(t.abs() < 1.0);
+//! ```
 use crate::matrix::{Matrix, SeriesOps};
 
 use crate::compute::stats::{gamma_cdf, mean, sample_variance};

src/compute/stats/mod.rs

@@ -1,3 +1,16 @@
+//! Statistical routines for matrices.
+//!
+//! Functions are grouped into submodules for descriptive statistics,
+//! correlations, probability distributions and basic inferential tests.
+//!
+//! ```
+//! use rustframe::compute::stats;
+//! use rustframe::matrix::Matrix;
+//!
+//! let m = Matrix::from_vec(vec![1.0, 2.0, 3.0, 4.0], 2, 2);
+//! let cov = stats::covariance(&m, &m);
+//! assert!((cov - 1.25).abs() < 1e-8);
+//! ```
 pub mod correlation;
 pub mod descriptive;
 pub mod distributions;

src/dataframe/df.rs

@@ -0,0 +1,659 @@
+use crate::frame::{Frame, RowIndex};
+use std::any::{Any, TypeId};
+use std::collections::HashMap;
+use std::fmt; // Import TypeId
+
+const DEFAULT_DISPLAY_ROWS: usize = 5;
+const DEFAULT_DISPLAY_COLS: usize = 10;
+
+// Trait to enable type-agnostic operations on Frame objects within DataFrame
+pub trait SubFrame: Send + Sync + fmt::Debug + Any {
+    fn rows(&self) -> usize;
+    fn get_value_as_string(&self, physical_row_idx: usize, col_name: &str) -> String;
+    fn clone_box(&self) -> Box<dyn SubFrame>;
+    fn delete_column_from_frame(&mut self, col_name: &str);
+    fn get_frame_cols(&self) -> usize; // Add a method to get the number of columns in the underlying frame
+
+    // Methods for downcasting to concrete types
+    fn as_any(&self) -> &dyn Any;
+    fn as_any_mut(&mut self) -> &mut dyn Any;
+}
+
+// Implement SubFrame for any Frame<T> that meets the requirements
+impl<T> SubFrame for Frame<T>
+where
+    T: Clone + PartialEq + fmt::Display + fmt::Debug + 'static + Send + Sync + Any,
+{
+    fn rows(&self) -> usize {
+        self.rows()
+    }
+
+    fn get_value_as_string(&self, physical_row_idx: usize, col_name: &str) -> String {
+        self.get_row(physical_row_idx).get(col_name).to_string()
+    }
+
+    fn clone_box(&self) -> Box<dyn SubFrame> {
+        Box::new(self.clone())
+    }
+
+    fn delete_column_from_frame(&mut self, col_name: &str) {
+        self.delete_column(col_name);
+    }
+
+    fn get_frame_cols(&self) -> usize {
+        self.cols()
+    }
+
+    fn as_any(&self) -> &dyn Any {
+        self
+    }
+
+    fn as_any_mut(&mut self) -> &mut dyn Any {
+        self
+    }
+}
+
+pub struct DataFrame {
+    frames_by_type: HashMap<TypeId, Box<dyn SubFrame>>, // Maps TypeId to the Frame holding columns of that type
+    column_to_type: HashMap<String, TypeId>,            // Maps column name to its TypeId
+    column_names: Vec<String>,
+    index: RowIndex,
+}
+
+impl DataFrame {
+    pub fn new() -> Self {
+        DataFrame {
+            frames_by_type: HashMap::new(),
+            column_to_type: HashMap::new(),
+            column_names: Vec::new(),
+            index: RowIndex::Range(0..0), // Initialize with an empty range index
+        }
+    }
+
+    /// Returns the number of rows in the DataFrame.
+    pub fn rows(&self) -> usize {
+        self.index.len()
+    }
+
+    /// Returns the number of columns in the DataFrame.
+    pub fn cols(&self) -> usize {
+        self.column_names.len()
+    }
+
+    /// Returns a reference to the vector of column names.
+    pub fn get_column_names(&self) -> &Vec<String> {
+        &self.column_names
+    }
+
+    /// Returns the number of internal Frame objects (one per unique data type).
+    pub fn num_internal_frames(&self) -> usize {
+        self.frames_by_type.len()
+    }
+
+    /// Returns a reference to a column of a specific type, if it exists.
+    pub fn get_column<T>(&self, col_name: &str) -> Option<&[T]>
+    where
+        T: Clone + PartialEq + fmt::Display + fmt::Debug + 'static + Send + Sync + Any,
+    {
+        let expected_type_id = TypeId::of::<T>();
+        if let Some(actual_type_id) = self.column_to_type.get(col_name) {
+            if *actual_type_id == expected_type_id {
+                if let Some(sub_frame_box) = self.frames_by_type.get(actual_type_id) {
+                    if let Some(frame) = sub_frame_box.as_any().downcast_ref::<Frame<T>>() {
+                        return Some(frame.column(col_name));
+                    }
+                }
+            }
+        }
+        None
+    }
+
+    /// Returns a HashMap representing a row, mapping column names to their string values.
+    pub fn get_row(&self, row_idx: usize) -> Option<HashMap<String, String>> {
+        if row_idx >= self.rows() {
+            return None;
+        }
+
+        let mut row_data = HashMap::new();
+        for col_name in &self.column_names {
+            if let Some(type_id) = self.column_to_type.get(col_name) {
+                if let Some(sub_frame_box) = self.frames_by_type.get(type_id) {
+                    let value = sub_frame_box.get_value_as_string(row_idx, col_name);
+                    row_data.insert(col_name.clone(), value);
+                }
+            }
+        }
+        Some(row_data)
+    }
+
+    pub fn add_column<T>(&mut self, col_name: &str, data: Vec<T>)
+    where
+        T: Clone + PartialEq + fmt::Display + fmt::Debug + 'static + Send + Sync + Any,
+    {
+        let type_id = TypeId::of::<T>();
+        let col_name_string = col_name.to_string();
+
+        // Check for duplicate column name across the entire DataFrame
+        if self.column_to_type.contains_key(&col_name_string) {
+            panic!(
+                "DataFrame::add_column: duplicate column name: '{}'",
+                col_name_string
+            );
+        }
+
+        // If this is the first column being added, set the DataFrame's index
+        if self.column_names.is_empty() {
+            self.index = RowIndex::Range(0..data.len());
+        } else {
+            // Ensure new column has the same number of rows as existing columns
+            if data.len() != self.index.len() {
+                panic!(
+                    "DataFrame::add_column: new column '{}' has {} rows, but existing columns have {} rows",
+                    col_name_string,
+                    data.len(),
+                    self.index.len()
+                );
+            }
+        }
+
+        // Check if a Frame of this type already exists
+        if let Some(sub_frame_box) = self.frames_by_type.get_mut(&type_id) {
+            // Downcast to the concrete Frame<T> and add the column
+            if let Some(frame) = sub_frame_box.as_any_mut().downcast_mut::<Frame<T>>() {
+                frame.add_column(col_name_string.clone(), data);
+            } else {
+                // This should ideally not happen if TypeId matches, but good for safety
+                panic!(
+                    "Type mismatch when downcasting existing SubFrame for TypeId {:?}",
+                    type_id
+                );
+            }
+        } else {
+            // No Frame of this type exists, create a new one
+            // The Frame::new constructor expects a Matrix and column names.
+            // We create a Matrix from a single column vector.
+            let new_frame = Frame::new(
+                crate::matrix::Matrix::from_cols(vec![data]),
+                vec![col_name_string.clone()],
+                Some(self.index.clone()), // Pass the DataFrame's index to the new Frame
+            );
+            self.frames_by_type.insert(type_id, Box::new(new_frame));
+        }
+
+        // Update column mappings and names
+        self.column_to_type.insert(col_name_string.clone(), type_id);
+        self.column_names.push(col_name_string);
+    }
+
+    /// Drops a column from the DataFrame.
+    /// Panics if the column does not exist.
+    pub fn drop_column(&mut self, col_name: &str) {
+        let col_name_string = col_name.to_string();
+
+        // 1. Get the TypeId associated with the column
+        let type_id = self
+            .column_to_type
+            .remove(&col_name_string)
+            .unwrap_or_else(|| {
+                panic!(
+                    "DataFrame::drop_column: column '{}' not found",
+                    col_name_string
+                );
+            });
+
+        // 2. Remove the column name from the ordered list
+        self.column_names.retain(|name| name != &col_name_string);
+
+        // 3. Find the Frame object and delete the column from it
+        if let Some(sub_frame_box) = self.frames_by_type.get_mut(&type_id) {
+            sub_frame_box.delete_column_from_frame(&col_name_string);
+
+            // 4. If the Frame object for this type becomes empty, remove it from frames_by_type
+            if sub_frame_box.get_frame_cols() == 0 {
+                self.frames_by_type.remove(&type_id);
+            }
+        } else {
+            // This should not happen if column_to_type was consistent
+            panic!(
+                "DataFrame::drop_column: internal error, no frame found for type_id {:?}",
+                type_id
+            );
+        }
+    }
+}
+
+impl fmt::Display for DataFrame {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        // Display column headers
+        for col_name in self.column_names.iter().take(DEFAULT_DISPLAY_COLS) {
+            write!(f, "{:<15}", col_name)?;
+        }
+        if self.column_names.len() > DEFAULT_DISPLAY_COLS {
+            write!(f, "...")?;
+        }
+        writeln!(f)?;
+
+        // Display data rows
+        let mut displayed_rows = 0;
+        for i in 0..self.index.len() {
+            if displayed_rows >= DEFAULT_DISPLAY_ROWS {
+                writeln!(f, "...")?;
+                break;
+            }
+            for col_name in self.column_names.iter().take(DEFAULT_DISPLAY_COLS) {
+                if let Some(type_id) = self.column_to_type.get(col_name) {
+                    if let Some(sub_frame_box) = self.frames_by_type.get(type_id) {
+                        write!(f, "{:<15}", sub_frame_box.get_value_as_string(i, col_name))?;
+                    } else {
+                        // This case indicates an inconsistency: column_to_type has an entry,
+                        // but frames_by_type doesn't have the corresponding Frame.
+                        write!(f, "{:<15}", "[ERROR]")?;
+                    }
+                } else {
+                    // This case indicates an inconsistency: column_names has an entry,
+                    // but column_to_type doesn't have the corresponding column.
+                    write!(f, "{:<15}", "[ERROR]")?;
+                }
+            }
+            if self.column_names.len() > DEFAULT_DISPLAY_COLS {
+                write!(f, "...")?;
+            }
+            writeln!(f)?;
+            displayed_rows += 1;
+        }
+        Ok(())
+    }
+}
+
+impl fmt::Debug for DataFrame {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.debug_struct("DataFrame")
+            .field("column_names", &self.column_names)
+            .field("index", &self.index)
+            .field("column_to_type", &self.column_to_type)
+            .field("frames_by_type", &self.frames_by_type)
+            .finish()
+    }
+}
+
+////////////////////////////////////////////////////////////////////////////////////////////////////
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::frame::Frame;
+    use crate::matrix::Matrix;
+
+    #[test]
+    fn test_dataframe_new() {
+        let df = DataFrame::new();
+        assert_eq!(df.rows(), 0);
+        assert_eq!(df.cols(), 0);
+        assert!(df.get_column_names().is_empty());
+        assert!(df.frames_by_type.is_empty());
+        assert!(df.column_to_type.is_empty());
+    }
+
+    #[test]
+    fn test_dataframe_add_column_initial() {
+        let mut df = DataFrame::new();
+        let data = vec![1, 2, 3];
+        df.add_column("col_int", data.clone());
+
+        assert_eq!(df.rows(), 3);
+        assert_eq!(df.cols(), 1);
+        assert_eq!(df.get_column_names(), &vec!["col_int".to_string()]);
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<i32>()));
+        assert_eq!(df.column_to_type.get("col_int"), Some(&TypeId::of::<i32>()));
+
+        // Verify the underlying frame
+        let sub_frame_box = df.frames_by_type.get(&TypeId::of::<i32>()).unwrap();
+        let frame = sub_frame_box.as_any().downcast_ref::<Frame<i32>>().unwrap();
+        assert_eq!(frame.rows(), 3);
+        assert_eq!(frame.cols(), 1);
+        assert_eq!(frame.columns(), &vec!["col_int".to_string()]);
+    }
+
+    #[test]
+    fn test_dataframe_add_column_same_type() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int1", vec![1, 2, 3]);
+        df.add_column("col_int2", vec![4, 5, 6]);
+
+        assert_eq!(df.rows(), 3);
+        assert_eq!(df.cols(), 2);
+        assert_eq!(
+            df.get_column_names(),
+            &vec!["col_int1".to_string(), "col_int2".to_string()]
+        );
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<i32>()));
+        assert_eq!(
+            df.column_to_type.get("col_int1"),
+            Some(&TypeId::of::<i32>())
+        );
+        assert_eq!(
+            df.column_to_type.get("col_int2"),
+            Some(&TypeId::of::<i32>())
+        );
+
+        // Verify the underlying frame
+        let sub_frame_box = df.frames_by_type.get(&TypeId::of::<i32>()).unwrap();
+        let frame = sub_frame_box.as_any().downcast_ref::<Frame<i32>>().unwrap();
+        assert_eq!(frame.rows(), 3);
+        assert_eq!(frame.cols(), 2);
+        assert_eq!(
+            frame.columns(),
+            &vec!["col_int1".to_string(), "col_int2".to_string()]
+        );
+    }
+
+    #[test]
+    fn test_dataframe_add_column_different_type() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int", vec![1, 2, 3]);
+        df.add_column("col_float", vec![1.1, 2.2, 3.3]);
+        df.add_column(
+            "col_string",
+            vec!["a".to_string(), "b".to_string(), "c".to_string()],
+        );
+
+        assert_eq!(df.rows(), 3);
+        assert_eq!(df.cols(), 3);
+        assert_eq!(
+            df.get_column_names(),
+            &vec![
+                "col_int".to_string(),
+                "col_float".to_string(),
+                "col_string".to_string()
+            ]
+        );
+
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<i32>()));
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<f64>()));
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<String>()));
+
+        assert_eq!(df.column_to_type.get("col_int"), Some(&TypeId::of::<i32>()));
+        assert_eq!(
+            df.column_to_type.get("col_float"),
+            Some(&TypeId::of::<f64>())
+        );
+        assert_eq!(
+            df.column_to_type.get("col_string"),
+            Some(&TypeId::of::<String>())
+        );
+
+        // Verify underlying frames
+        let int_frame = df
+            .frames_by_type
+            .get(&TypeId::of::<i32>())
+            .unwrap()
+            .as_any()
+            .downcast_ref::<Frame<i32>>()
+            .unwrap();
+        assert_eq!(int_frame.columns(), &vec!["col_int".to_string()]);
+
+        let float_frame = df
+            .frames_by_type
+            .get(&TypeId::of::<f64>())
+            .unwrap()
+            .as_any()
+            .downcast_ref::<Frame<f64>>()
+            .unwrap();
+        assert_eq!(float_frame.columns(), &vec!["col_float".to_string()]);
+
+        let string_frame = df
+            .frames_by_type
+            .get(&TypeId::of::<String>())
+            .unwrap()
+            .as_any()
+            .downcast_ref::<Frame<String>>()
+            .unwrap();
+        assert_eq!(string_frame.columns(), &vec!["col_string".to_string()]);
+    }
+
+    #[test]
+    fn test_dataframe_get_column() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int", vec![1, 2, 3]);
+        df.add_column("col_float", vec![1.1, 2.2, 3.3]);
+        df.add_column(
+            "col_string",
+            vec!["a".to_string(), "b".to_string(), "c".to_string()],
+        );
+
+        // Test getting existing columns with correct type
+        assert_eq!(
+            df.get_column::<i32>("col_int").unwrap(),
+            vec![1, 2, 3].as_slice()
+        );
+        assert_eq!(
+            df.get_column::<f64>("col_float").unwrap(),
+            vec![1.1, 2.2, 3.3].as_slice()
+        );
+        assert_eq!(
+            df.get_column::<String>("col_string").unwrap(),
+            vec!["a".to_string(), "b".to_string(), "c".to_string()].as_slice()
+        );
+
+        // Test getting non-existent column
+        assert_eq!(df.get_column::<i32>("non_existent"), None);
+
+        // Test getting existing column with incorrect type
+        assert_eq!(df.get_column::<f64>("col_int"), None);
+        assert_eq!(df.get_column::<i32>("col_float"), None);
+    }
+
+    #[test]
+    fn test_dataframe_get_row() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int", vec![1, 2, 3]);
+        df.add_column("col_float", vec![1.1, 2.2, 3.3]);
+        df.add_column(
+            "col_string",
+            vec!["a".to_string(), "b".to_string(), "c".to_string()],
+        );
+
+        // Test getting an existing row
+        let row0 = df.get_row(0).unwrap();
+        assert_eq!(row0.get("col_int"), Some(&"1".to_string()));
+        assert_eq!(row0.get("col_float"), Some(&"1.1".to_string()));
+        assert_eq!(row0.get("col_string"), Some(&"a".to_string()));
+
+        let row1 = df.get_row(1).unwrap();
+        assert_eq!(row1.get("col_int"), Some(&"2".to_string()));
+        assert_eq!(row1.get("col_float"), Some(&"2.2".to_string()));
+        assert_eq!(row1.get("col_string"), Some(&"b".to_string()));
+
+        // Test getting an out-of-bounds row
+        assert_eq!(df.get_row(3), None);
+    }
+
+    #[test]
+    #[should_panic(expected = "DataFrame::add_column: duplicate column name: 'col_int'")]
+    fn test_dataframe_add_column_duplicate_name() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int", vec![1, 2, 3]);
+        df.add_column("col_int", vec![4, 5, 6]);
+    }
+
+    #[test]
+    #[should_panic(
+        expected = "DataFrame::add_column: new column 'col_int2' has 2 rows, but existing columns have 3 rows"
+    )]
+    fn test_dataframe_add_column_mismatched_rows() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int1", vec![1, 2, 3]);
+        df.add_column("col_int2", vec![4, 5]);
+    }
+
+    #[test]
+    fn test_dataframe_display() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int", vec![1, 2, 3, 4, 5, 6]);
+        df.add_column("col_float", vec![1.1, 2.2, 3.3, 4.4, 5.5, 6.6]);
+        df.add_column(
+            "col_string",
+            vec![
+                "a".to_string(),
+                "b".to_string(),
+                "c".to_string(),
+                "d".to_string(),
+                "e".to_string(),
+                "f".to_string(),
+            ],
+        );
+
+        let expected_output = "\
+col_int        col_float      col_string     
+1              1.1            a              
+2              2.2            b              
+3              3.3            c              
+4              4.4            d              
+5              5.5            e              
+...
+";
+        assert_eq!(format!("{}", df), expected_output);
+    }
+
+    #[test]
+    fn test_dataframe_debug() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int", vec![1, 2, 3]);
+        df.add_column("col_float", vec![1.1, 2.2, 3.3]);
+
+        let debug_output = format!("{:?}", df);
+        assert!(debug_output.contains("DataFrame {"));
+        assert!(debug_output.contains("column_names: [\"col_int\", \"col_float\"]"));
+        assert!(debug_output.contains("index: Range(0..3)"));
+        assert!(debug_output.contains("column_to_type: {"));
+        assert!(debug_output.contains("frames_by_type: {"));
+    }
+
+    #[test]
+    fn test_dataframe_drop_column_single_type() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int1", vec![1, 2, 3]);
+        df.add_column("col_int2", vec![4, 5, 6]);
+        df.add_column("col_float", vec![1.1, 2.2, 3.3]);
+
+        assert_eq!(df.cols(), 3);
+        assert_eq!(
+            df.get_column_names(),
+            &vec![
+                "col_int1".to_string(),
+                "col_int2".to_string(),
+                "col_float".to_string()
+            ]
+        );
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<i32>()));
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<f64>()));
+
+        df.drop_column("col_int1");
+
+        assert_eq!(df.cols(), 2);
+        assert_eq!(
+            df.get_column_names(),
+            &vec!["col_int2".to_string(), "col_float".to_string()]
+        );
+        assert!(df.column_to_type.get("col_int1").is_none());
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<i32>())); // Frame<i32> should still exist
+        let int_frame = df
+            .frames_by_type
+            .get(&TypeId::of::<i32>())
+            .unwrap()
+            .as_any()
+            .downcast_ref::<Frame<i32>>()
+            .unwrap();
+        assert_eq!(int_frame.columns(), &vec!["col_int2".to_string()]);
+
+        df.drop_column("col_int2");
+
+        assert_eq!(df.cols(), 1);
+        assert_eq!(df.get_column_names(), &vec!["col_float".to_string()]);
+        assert!(df.column_to_type.get("col_int2").is_none());
+        assert!(!df.frames_by_type.contains_key(&TypeId::of::<i32>())); // Frame<i32> should be removed
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<f64>()));
+    }
+
+    #[test]
+    fn test_dataframe_drop_column_mixed_types() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int", vec![1, 2, 3]);
+        df.add_column("col_float", vec![1.1, 2.2, 3.3]);
+        df.add_column(
+            "col_string",
+            vec!["a".to_string(), "b".to_string(), "c".to_string()],
+        );
+
+        assert_eq!(df.cols(), 3);
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<i32>()));
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<f64>()));
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<String>()));
+
+        df.drop_column("col_float");
+
+        assert_eq!(df.cols(), 2);
+        assert_eq!(
+            df.get_column_names(),
+            &vec!["col_int".to_string(), "col_string".to_string()]
+        );
+        assert!(df.column_to_type.get("col_float").is_none());
+        assert!(!df.frames_by_type.contains_key(&TypeId::of::<f64>())); // Frame<f64> should be removed
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<i32>()));
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<String>()));
+
+        df.drop_column("col_int");
+        df.drop_column("col_string");
+
+        assert_eq!(df.cols(), 0);
+        assert!(df.get_column_names().is_empty());
+        assert!(df.frames_by_type.is_empty());
+        assert!(df.column_to_type.is_empty());
+    }
+
+    #[test]
+    #[should_panic(expected = "DataFrame::drop_column: column 'non_existent' not found")]
+    fn test_dataframe_drop_column_non_existent() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int", vec![1, 2, 3]);
+        df.drop_column("non_existent");
+    }
+
+    #[test]
+    fn test_dataframe_add_column_reuses_existing_frame() {
+        let mut df = DataFrame::new();
+        df.add_column("col_int1", vec![1, 2, 3]);
+        df.add_column("col_float1", vec![1.1, 2.2, 3.3]);
+
+        // Initially, there should be two frames (one for i32, one for f64)
+        assert_eq!(df.frames_by_type.len(), 2);
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<i32>()));
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<f64>()));
+
+        // Add another integer column
+        df.add_column("col_int2", vec![4, 5, 6]);
+
+        // The number of frames should still be 2, as the existing i32 frame should be reused
+        assert_eq!(df.frames_by_type.len(), 2);
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<i32>()));
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<f64>()));
+
+        // Verify the i32 frame now contains both integer columns
+        let int_frame = df.frames_by_type.get(&TypeId::of::<i32>()).unwrap().as_any().downcast_ref::<Frame<i32>>().unwrap();
+        assert_eq!(int_frame.columns(), &vec!["col_int1".to_string(), "col_int2".to_string()]);
+        assert_eq!(int_frame.cols(), 2);
+
+        // Add another float column
+        df.add_column("col_float2", vec![4.4, 5.5, 6.6]);
+
+        // The number of frames should still be 2, as the existing f64 frame should be reused
+        assert_eq!(df.frames_by_type.len(), 2);
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<i32>()));
+        assert!(df.frames_by_type.contains_key(&TypeId::of::<f64>()));
+
+        // Verify the f64 frame now contains both float columns
+        let float_frame = df.frames_by_type.get(&TypeId::of::<f64>()).unwrap().as_any().downcast_ref::<Frame<f64>>().unwrap();
+        assert_eq!(float_frame.columns(), &vec!["col_float1".to_string(), "col_float2".to_string()]);
+        assert_eq!(float_frame.cols(), 2);
+    }
+}

src/dataframe/mod.rs

@@ -0,0 +1,4 @@
+//! This module provides the DataFrame structure for handling tabular data with mixed types.
+pub mod df;
+
+pub use df::{DataFrame, SubFrame};

src/frame/base.rs

@@ -1,3 +1,19 @@
+//! Core data-frame structures such as [`Frame`] and [`RowIndex`].
+//!
+//! The [`Frame`] type stores column-labelled data with an optional row index
+//! and builds upon the [`crate::matrix::Matrix`] type.
+//!
+//! # Examples
+//!
+//! ```
+//! use rustframe::frame::{Frame, RowIndex};
+//! use rustframe::matrix::Matrix;
+//!
+//! let data = Matrix::from_cols(vec![vec![1, 2], vec![3, 4]]);
+//! let frame = Frame::new(data, vec!["L", "R"], Some(RowIndex::Int(vec![10, 20])));
+//! assert_eq!(frame.columns(), &["L", "R"]);
+//! assert_eq!(frame.index(), &RowIndex::Int(vec![10, 20]));
+//! ```
 use crate::matrix::Matrix;
 use chrono::NaiveDate;
 use std::collections::HashMap;
@@ -316,7 +332,7 @@ impl<T: Clone + PartialEq> Frame<T> {
         )
     }
 
-    /// Returns an immutable slice of the specified column's data.
+    /// Returns an immutable slice of the specified column's data by name.
     /// Panics if the column name is not found.
     pub fn column(&self, name: &str) -> &[T] {
         let idx = self
@@ -325,7 +341,13 @@ impl<T: Clone + PartialEq> Frame<T> {
         self.matrix.column(idx)
     }
 
-    /// Returns a mutable slice of the specified column's data.
+    /// Returns an immutable slice of the specified column's data by its physical index.
+    /// Panics if the index is out of bounds.
+    pub fn column_by_physical_idx(&self, idx: usize) -> &[T] {
+        self.matrix.column(idx)
+    }
+
+    /// Returns a mutable slice of the specified column's data by name.
     /// Panics if the column name is not found.
     pub fn column_mut(&mut self, name: &str) -> &mut [T] {
         let idx = self
@@ -334,6 +356,12 @@ impl<T: Clone + PartialEq> Frame<T> {
         self.matrix.column_mut(idx)
     }
 
+    /// Returns a mutable slice of the specified column's data by its physical index.
+    /// Panics if the index is out of bounds.
+    pub fn column_mut_by_physical_idx(&mut self, idx: usize) -> &mut [T] {
+        self.matrix.column_mut(idx)
+    }
+
     // Row access methods
 
     /// Returns an immutable view of the row for the given integer key.

src/frame/mod.rs

@@ -1,3 +1,21 @@
+//! High-level interface for working with columnar data and row indices.
+//!
+//! The [`Frame`](crate::frame::Frame) type combines a matrix with column labels and a typed row
+//! index, similar to data frames in other data-analysis libraries.
+//!
+//! # Examples
+//!
+//! ```
+//! use rustframe::frame::{Frame, RowIndex};
+//! use rustframe::matrix::Matrix;
+//!
+//! // Build a frame from two columns labelled "A" and "B".
+//! let data = Matrix::from_cols(vec![vec![1.0, 2.0], vec![3.0, 4.0]]);
+//! let frame = Frame::new(data, vec!["A", "B"], None);
+//!
+//! assert_eq!(frame["A"], vec![1.0, 2.0]);
+//! assert_eq!(frame.index(), &RowIndex::Range(0..2));
+//! ```
 pub mod base;
 pub mod ops;
 

src/frame/ops.rs

@@ -1,3 +1,16 @@
+//! Trait implementations that allow [`Frame`] to reuse matrix operations.
+//!
+//! These modules forward numeric and boolean aggregation methods from the
+//! underlying [`Matrix`](crate::matrix::Matrix) type so that they can be called
+//! directly on a [`Frame`].
+//!
+//! ```
+//! use rustframe::frame::Frame;
+//! use rustframe::matrix::{Matrix, SeriesOps};
+//!
+//! let frame = Frame::new(Matrix::from_cols(vec![vec![1.0, 2.0]]), vec!["A"], None);
+//! assert_eq!(frame.sum_vertical(), vec![3.0]);
+//! ```
 use crate::frame::Frame;
 use crate::matrix::{Axis, BoolMatrix, BoolOps, FloatMatrix, SeriesOps};
 

src/lib.rs

@@ -1,5 +1,8 @@
 #![doc = include_str!("../README.md")]
 
+/// Documentation for the [`crate::dataframe`] module.
+pub mod dataframe;
+
 /// Documentation for the [`crate::matrix`] module.
 pub mod matrix;
 

src/matrix/boolops.rs

@@ -1,3 +1,14 @@
+//! Logical reductions for boolean matrices.
+//!
+//! The [`BoolOps`] trait mirrors common boolean aggregations such as `any` and
+//! `all` over rows or columns of a [`BoolMatrix`].
+//!
+//! ```
+//! use rustframe::matrix::{BoolMatrix, BoolOps};
+//!
+//! let m = BoolMatrix::from_vec(vec![true, false], 2, 1);
+//! assert!(m.any());
+//! ```
 use crate::matrix::{Axis, BoolMatrix};
 
 /// Boolean operations on `Matrix<bool>`

src/matrix/mod.rs

@@ -1,3 +1,18 @@
+//! Core matrix types and operations.
+//!
+//! The [`Matrix`](crate::matrix::Matrix) struct provides a simple column‑major 2D array with a
+//! suite of numeric helpers. Additional traits like [`SeriesOps`](crate::matrix::SeriesOps) and
+//! [`BoolOps`](crate::matrix::BoolOps) extend functionality for common statistics and logical reductions.
+//!
+//! # Examples
+//!
+//! ```
+//! use rustframe::matrix::Matrix;
+//!
+//! let m = Matrix::from_cols(vec![vec![1, 2], vec![3, 4]]);
+//! assert_eq!(m.shape(), (2, 2));
+//! assert_eq!(m[(0,1)], 3);
+//! ```
 pub mod boolops;
 pub mod mat;
 pub mod seriesops;

src/matrix/seriesops.rs

@@ -1,3 +1,14 @@
+//! Numeric reductions and transformations over matrix axes.
+//!
+//! [`SeriesOps`] provides methods like [`SeriesOps::sum_vertical`] or
+//! [`SeriesOps::map`] that operate on [`FloatMatrix`] values.
+//!
+//! ```
+//! use rustframe::matrix::{Matrix, SeriesOps};
+//!
+//! let m = Matrix::from_cols(vec![vec![1.0, 2.0], vec![3.0, 4.0]]);
+//! assert_eq!(m.sum_horizontal(), vec![4.0, 6.0]);
+//! ```
 use crate::matrix::{Axis, BoolMatrix, FloatMatrix};
 
 /// "Series-like" helpers that work along a single axis.

src/random/crypto.rs

@@ -1,3 +1,13 @@
+//! Cryptographically secure random number generator.
+//!
+//! On Unix systems this reads from `/dev/urandom`; on Windows it uses the
+//! system's preferred CNG provider.
+//!
+//! ```
+//! use rustframe::random::{crypto_rng, Rng};
+//! let mut rng = crypto_rng();
+//! let _v = rng.next_u64();
+//! ```
 #[cfg(unix)]
 use std::{fs::File, io::Read};
 

src/random/mod.rs

@@ -1,3 +1,18 @@
+//! Random number generation utilities.
+//!
+//! Provides both a simple pseudo-random generator [`Prng`](crate::random::Prng) and a
+//! cryptographically secure alternative [`CryptoRng`](crate::random::CryptoRng). The
+//! [`SliceRandom`](crate::random::SliceRandom) trait offers shuffling of slices using any RNG
+//! implementing [`Rng`](crate::random::Rng).
+//!
+//! ```
+//! use rustframe::random::{rng, SliceRandom};
+//!
+//! let mut rng = rng();
+//! let mut data = [1, 2, 3, 4];
+//! data.shuffle(&mut rng);
+//! assert_eq!(data.len(), 4);
+//! ```
 pub mod crypto;
 pub mod prng;
 pub mod random_core;

src/random/prng.rs

@@ -1,3 +1,11 @@
+//! A tiny XorShift64-based pseudo random number generator.
+//!
+//! ```
+//! use rustframe::random::{rng, Rng};
+//! let mut rng = rng();
+//! let x = rng.next_u64();
+//! assert!(x >= 0);
+//! ```
 use std::time::{SystemTime, UNIX_EPOCH};
 
 use crate::random::Rng;

src/random/random_core.rs

@@ -1,3 +1,11 @@
+//! Core traits for random number generators and sampling ranges.
+//!
+//! ```
+//! use rustframe::random::{rng, Rng};
+//! let mut r = rng();
+//! let value: f64 = r.random_range(0.0..1.0);
+//! assert!(value >= 0.0 && value < 1.0);
+//! ```
 use std::f64::consts::PI;
 use std::ops::Range;
 

src/random/seq.rs

@@ -1,3 +1,11 @@
+//! Extensions for shuffling slices with a random number generator.
+//!
+//! ```
+//! use rustframe::random::{rng, SliceRandom};
+//! let mut data = [1, 2, 3];
+//! data.shuffle(&mut rng());
+//! assert_eq!(data.len(), 3);
+//! ```
 use crate::random::Rng;
 
 /// Trait for randomizing slices.

src/utils/dateutils/dates.rs

@@ -1,3 +1,10 @@
+//! Generation and manipulation of calendar date sequences.
+//!
+//! ```
+//! use rustframe::utils::dateutils::dates::{DateFreq, DatesList};
+//! let list = DatesList::new("2024-01-01".into(), "2024-01-03".into(), DateFreq::Daily);
+//! assert_eq!(list.count().unwrap(), 3);
+//! ```
 use chrono::{Datelike, Duration, NaiveDate, Weekday};
 use std::collections::HashMap;
 use std::error::Error;

src/utils/dateutils/mod.rs

@@ -1,3 +1,13 @@
+//! Generators for sequences of calendar and business dates.
+//!
+//! See [`dates`] for all-day calendars and [`bdates`] for business-day aware
+//! variants.
+//!
+//! ```
+//! use rustframe::utils::dateutils::{DatesList, DateFreq};
+//! let list = DatesList::new("2024-01-01".into(), "2024-01-02".into(), DateFreq::Daily);
+//! assert_eq!(list.count().unwrap(), 2);
+//! ```
 pub mod bdates;
 pub mod dates;
 

src/utils/mod.rs

@@ -1,3 +1,14 @@
+//! Assorted helper utilities.
+//!
+//! Currently this module exposes date generation utilities in [`dateutils`](crate::utils::dateutils),
+//! including calendar and business date sequences.
+//!
+//! ```
+//! use rustframe::utils::DatesList;
+//! use rustframe::utils::DateFreq;
+//! let dates = DatesList::new("2024-01-01".into(), "2024-01-03".into(), DateFreq::Daily);
+//! assert_eq!(dates.count().unwrap(), 3);
+//! ```
 pub mod dateutils;
 
 pub use dateutils::{BDateFreq, BDatesGenerator, BDatesList};