mfamador / ml-test

Machine learning tests

Machine Learning experiments

Install requirements

pip3.6 install -r requirements.txt 

Jupyter notebooks to analyse the dataset and create the model for the different scenarios:

multi-label-text-classification.ipynb

time-series-prediction-LSTM.ipynb.ipynb

Train and save the model

./1-create-model.sh

or

python3.6 src/modeltrainer/model_trainer.py -i receipt_data.csv 

Service to predict retailer name from OCR raw data

Run predictor service

Run gunicorn server locally:

PYTHONPATH=src/retailerpredictor gunicorn -c gunicorn.cfg app:app 

Run gunicorn server with docker:

./2-build-docker-image.sh
./3-start-docker-container.sh

Test a prediction

curl 'http://localhost:8080/predict?retailer=boots%20uk'