DjangoAPI SOLUTION

Solution

1. Download DjangoAPI.zip, unpack;
2. Go to the \DjangoAPI directory;
3. Start the server using the command python Manage.py runserver;
4. Now instructions for the specified result of the json file to be sent to the server.
   • Use the HTTP request POST http://127.0.0.1:8000/match_products;
   • Receive a json response with predictions in the body;

Model training

1. Download TwoModels.py, arcNet.py, Ensemble.py, in the same directory agora_hack_products.json
2. ExtraTreesClassifier[Classic ML solution] and Tokenizator[Data usage and mapping]:
   1. Run the command python TwoModels.py train - as a result, models and additional files will be saved;
   2. Running the command python TwoModels.py token will show the result of Tokenizator's work on all data;
   3. Running the command python TwoModels.py test will show the result of ExtraTreeClassifier running on all data;
3. ArcNet [Neural network solution]:
   1. Running the command python arcNet.py will output the expected parameters as input to the py file;
   2. Running the command python arcNet.py train [true/false] [true/false] [true/false] [true/false] [true/false] will train the models and save them to files upon request, also saving additional files;
   3. Running the command python arcNet.py test [true/false] will check the operation of the model on test data, and also save the head of the model (KNN-1) if necessary.
   4. Running the command python arcNet.py check will show the accuracy and running time on all data;