Maslov Victor 972201

"How To"

Without Docker

1. Download repository
2. Install spaceshiptitanic-0.1.0-py3-none-any.whl by pip or conda
3. Run command 'flask app.py' from root repo folder

Train model using CLI

Run command 'python model.py train --dataset=/path/to/dataset' from root repo folder

Get predictions using CLI

Run command 'python model.py predict --dataset=/path/to/dataset' from root repo folder

Train model using API

Send request POST to http://127.0.0.1:5000/train with body { "dataset_path": "/path/to/dataset" }

Get predictions using API

Send request POST to http://127.0.0.1:5000/predict with body { "dataset_path": "/path/to/dataset" }

With Docker

1. Download repository
2. Run command 'docker compose up -d' from root repo folder
3. Run command 'flask app.py' from 'app' folder in container

Train model using CLI

Run command 'python model.py train --dataset=/path/to/dataset' from 'app' folder in container

Get predictions using CLI

Run command 'python model.py predict --dataset=/path/to/dataset' from 'app' folder in container

Train model using API

Send request POST to http://127.0.0.1:5000/train in container with body { "dataset_path": "/path/to/dataset/from/app/folder" }

Get predictions using API

Send request POST to http://127.0.0.1:5000/predict in container with body { "dataset_path": "/path/to/dataset/from/app/folder" }

Utilizided resources

https://education.yandex.ru/handbook/ml https://catboost.ai/ https://forecastegy.com/posts/catboost-hyperparameter-tuning-guide-with-optuna/ https://scikit-learn.org/stable/ https://docs.docker.com/reference/cli/docker/image/build/ https://flask.palletsprojects.com/en/3.0.x/