Linear-Regression

This MCP server provides an end-to-end workflow for training a Linear Regression model using Claude and the MCP protocol. Users upload a CSV dataset, Claude handles data preprocessing (including column inspection and detection of categorical features), optional label encoding for categorical columns, and then trains a linear regression model while computing the RMSE as an evaluation metric. The server exposes tools to upload data, inspect columns, identify categorical features, encode categorical data, and kick off the model training pipeline with a specified target column. You can invoke these tools programmatically or via Claude Desktop integration to iterate on data preparation and model training without leaving the Claude environment.

Prerequisites:\n- Python environment (with uv installed) and access to the uv project manager.\n- A clone of the Linear-Regression-MCP repository.\n- Claude Desktop installed and configured for MCP servers.\n\nInstallation steps:\n1. Clone the repository:\nbash\ngit clone https://github.com/HeetVekariya/Linear-Regression-MCP\ncd Linear-Regression-MCP\n\n2. Install uv (Python package/project manager):\n- Follow instructions from the uv official documentation and install it on your system. This typically involves downloading the installer and adding the binary to your PATH.\n3. Install dependencies with uv:\nbash\nuv sync\n\n4. Configure Claude Desktop to connect to the MCP server (see README for details):\n- Open Claude Desktop configuration and set the mcpServers entry to point to your local uv-based server path, e.g.\njson\n{\n "mcpServers": {\n "linear-regression": {\n "command": "ABSOLUTE/PATH/TO/.local/bin/uv",\n "args": [\n "--directory",\n "ABSOLUTE/PATH/TO/YOUR-LINEAR-REGRESSION-REPO", \n "run",\n "server.py"\n ] \n }\n }\n}\n

Tips and common issues:\n- Ensure the absolute paths in the Claude config are correct to the uv executable and the Linear Regression MCP repo.\n- The dataset should have a clearly defined target column for the model to predict. Use the provided tools to inspect and preprocess data before training.\n- If you encounter encoding issues with categorical features, run label_encode_categorical_columns() before train_linear_regression_model(output_column).\n- RMSE is the primary evaluation metric used by train_linear_regression_model; ensure your dataset contains sufficient variance and a proper split for meaningful RMSE computation.