How daksh3008/stock_predictor works
Project type and purpose
• This is a Python-based machine learning pipeline built around an LSTM model.
• It ingests raw data, preprocesses it, trains an LSTM network, and produces predictions.
• There is no user interface or external HTTP API—this is a standalone data-science tool that you run via the command line or as a script.
File structure
• Top-level files:
• requirements.txt – lists Python dependencies.
• README.md – project overview (currently empty).
• src/main.py – the orchestration entry point.
• src/data: data ingestion and preprocessing
• fetch_data.py – pulls raw time-series data (from file, database, or external API).
• preprocess.py – cleans and transforms raw data into model-ready format.
• src/model: model definition, training, and prediction
• lstm_model.py – defines the LSTM network architecture.
• train.py – training loop, hyperparameters, model checkpoints.
• predict.py – loads a trained checkpoint and generates forecasts.
• src/utils: configuration and helper functions
• config.py – central place for hyperparameters, file paths, model settings.
• helpers.py – common routines (e.g., logging, data splitting).
• plot_utils.py – visualization helpers for loss curves and predictions.
README for additional insights
• Since it’s blank, rely on file names and inferred purpose.