manojmanivannan/ml_pytorch

Anomaly Detection (for Production)

Model generated and training using the notebook Anomaly Detection.

Assuming stock_data_generator used, the model is saved in the directory model, which is also provided. When using docker compose this model is loaded into the torchserve container which serves this model.

The Streamlit container, simply generates a new dataset from stock_data_generator and performs prediction using a time window of 24, which was used in the original model.

Note

Simply run docker-compose up --build if you wan to use the default model in this repo.