Description

FIFA19 is video game series about football produced by EA SPORTS . This game provide estimation of player's attribute,that correspond to technical abilities like dribbling,shooting etc or non technical thing like international reputation , work rate.Based on some weighting on attribute ,that depend on player's position , overall rating of player calculated .On this project , i want to predict overall rating of player based on some technical and non technical abilities that provided. The data can be downloaded from https://www.kaggle.com/karangadiya/fifa19. Information about attribute can be seen on https://fifauteam.com/fifa-19-attributes-guide/#22 . Position of football player can be seen on https://sofifa.com/calculator

File in repository :

• train.py : to train the best model i choose , after run it save model on model_chosen.bin
• predict.py : to deploy web service locally
• notebook.ipynb : Model selection process
• predict-test.py : to try web service that deployed locally
• deploy-test.py : to try web service that deployed on pythonanywhere
• prep.py : Module that needed on predict.py
• Dockerfile : to running the service on docker
• Summary-model.xlsx : summary of model selection process
• Data.csv : Data i used to train and test on this project

How to run project

Starter

1. Download project
2. Install dependencies with pipenv install

To run notebook.ipynb

1. Activate virtual environment in directory by pipenv shell and open jupyter notebook
2. Open notebook.ipynb
3. Run first section 1(Table of contents ) to section 5(EDA) .
4. For section 6.1 and 6.2 , you can run 6.2 first than 6.1 , but in each subsection of 6.1 and 6.2 , must run sequentially
5. For section 6.2 to 9.4 , run first section between 6.2 and 6.3.No need to run sequentially in section 6.2 to 9.4 , but need to run sequentially in each section.

To run train.py

1. Activate virtual environment in directory by pipenv shell
2. run python train.py , it will save model_chosen.bin in directory

Deploy locally using docker

1. run docker image using docker run -it --rm -p 9696:9696 zoomcamp-project
2. run python predict-test.py on another command prompt
3. Player data that specified in predict-test.py can modified if you want to try another player

Deploy the service on cloud

I deploy this project on pythonanywhere

Thank you for reviewing my project