DataFrame - C++ Data Analysis Library

A lightweight, high-performance C++ library for data manipulation and analysis, inspired by pandas.

Features

• Fast and Efficient: Built with performance in mind using modern C++
• DataFrame Operations: Select, filter, head, and more
• Column-based Storage: Efficient columnar data structure
• Groupby Operations: Group data and perform aggregations (count, sum, mean)
• CSV Support: Read data from CSV files or strings
• Method Chaining: Fluent API for data pipelines

Quick Start

#include "DataFrame.h"
#include "CsvReader.h"
#include <iostream>

int main() {
    // Read CSV data
    std::string csv =
        "price,volume,signal\n"
        "101.5,2000,1\n"
        "102.3,1800,0\n"
        "99.8,2500,-1\n";

    DataFrame df = readCSVString(csv);

    // Select columns
    auto view = df.select({ "price", "signal" });

    // Filter data
    auto filtered = view.filter("signal", [](double x) { return x != 0; });

    // Group by and aggregate
    auto result = df.groupby("signal").mean("price");

    // Print results
    std::cout << result << "\n";

    return 0;
}

Core Components

DataFrame

Main data structure that holds columnar data with named columns.

Key Methods:

• select(columns) - Select specific columns
• head(n) - Get first n rows
• groupby(column) - Group by column values
• numRows(), numCols() - Get dimensions
• operator[] - Access columns by name or index

DataFrameView

Lightweight view into a DataFrame without copying data.

Key Methods:

• select(columns) - Select columns from view
• filter(column, predicate) - Filter rows based on condition
• head(n) - Get first n rows
• at(row, col) - Access specific cell

GroupedDataFrame

Result of groupby operation, enables aggregations.

Aggregation Methods:

• count() - Count rows per group
• sum(column) - Sum values per group
• mean(column) - Calculate mean per group

Column

Represents a single column of double values.

Methods:

• sum(), mean(), min(), max(), median() - Aggregation functions
• operator[] - Access values by index
• size() - Get number of elements

Building the Project

Requirements

• C++20 or later
• Visual Studio 2019/2022 (or any C++20 compatible compiler)

Build with Visual Studio

1. Open DataFrame.sln
2. Select your desired configuration (Debug/Release)
3. Build the solution (F7)

Build with CMake (if needed)

mkdir build
cd build
cmake ..
cmake --build .

Performance

Benchmark results on 1,000,000 rows:

Operation	Time
select()	~4.5 ms
select + head()	~5.5 ms
select + filter()	~53 ms
groupby + count()	~179 ms
groupby + mean()	~228 ms
Chained pipeline	~207 ms

Project Structure

DataFrame/
??? Column.h              # Column implementation
??? DataFrame.h           # DataFrame class
??? DataFrame.cpp         # DataFrame implementation
??? DataFrameView.h       # View implementation
??? DataFrameView.cpp     # View implementation
??? GroupedDataFrame.h    # Groupby operations
??? GroupedDataFrame.cpp  # Groupby implementation
??? CsvReader.h           # CSV parsing
??? CsvReader.cpp         # CSV implementation
??? ColumnIO.h            # Print operators
??? Benchmark.h           # Benchmarking utilities
??? main.cpp              # Example usage

Example: Data Pipeline

// Read large dataset
DataFrame df = make_big_df(1'000'000);

// Chain operations
auto result = df
    .select({ "price", "signal" })
    .filter("signal", [](double x) { return x != 0; })
    .groupby("signal")
    .mean("price");

// Print formatted results
std::cout << result << "\n";

CSV Reading

// From string
std::string csv = "col1,col2\n1.0,2.0\n3.0,4.0\n";
DataFrame df = readCSVString(csv);

// From file
DataFrame df2 = readCSV("data.csv");

Printing DataFrames

The library includes formatted printing for all data structures:

std::cout << df << "\n";           // Print DataFrame
std::cout << view << "\n";         // Print DataFrameView
std::cout << df["price"] << "\n";  // Print Column

Output format:

       price        volume        signal
------------  ------------  ------------
       101.5          2000             1
       102.3          1800             0
        99.8          2500            -1

License

This project is open source. Feel free to use and modify as needed.

Contributing

Contributions are welcome! Please feel free to submit issues or pull requests.

Author

Created as a learning project to explore modern C++ and data structures.