MLOps Fifa players predict value

Assets in repo

• requirements.txt: View requirements.txt
• Futbol_Predictions_Export_Model.ipynb: View Futbol_Predictions_Export_Model.ipynb EDA and Export model
• data_fifa_players.json: View data Data scraping
• model.joblib: View model.joblib Exported Model
• app.py: View app.py Flask api
• mlib.py: View mlib.py Model Handling Library
• cli.py: View cli.py Console predict
• test_mlib.py: View test_mlib.py Unit test
• utilscli.py: View utilscli.py Utility Belt
• Dockerfile: View Dockerfile
• Makefile: View Makefile
• main.yml: View main.yml Github Actions
• webscrapi_fifa_players: View webscrapi_fifa_players Scrapy proyect

CLI Tools

There are two cli tools. First, the main cli.py is the endpoint that serves out predictions. To predict the height of an MLB player you use the following: (if you don't run on Windows delete the backslash "\")

$ python ./cli.py --skills '[{\"overallrating\": 75, \"potencial\": 76, \"height\": 174, \"weight\": 67, \"Age\": 26, \"Contract Length\": 2024, \"Ball Control\": 78.0, \"Dribbling\": 81.0, \"Marking\": 32.0, \"Slide Tackle\": 26.0, \"Stand Tackle\": 30.0, \"Aggression\": 61.0, \"Reactions\": 73.0, \"Att. Position\": 72.0, \"Interceptions\": 49.0, \"Vision\": 70.0, \"Composure\": 79.0, \"Crossing\": 66.0, \"Short Pass\": 71.0, \"Long Pass\": 63.0, \"Acceleration\": 83.0, \"Stamina\": 76.0, \"Strength\": 55.0, \"Balance\": 86.0, \"Sprint Speed\": 75.0, \"Agility\": 89.0, \"Jumping\": 76.0, \"Heading\": 50.0, \"Shot Power\": 71.0, \"Finishing\": 76.0, \"Long Shots\": 73.0, \"Curve\": 73.0, \"FK Acc.\": 69.0, \"Penalties\": 72.0, \"Volleys\": 70.0, \"GK Positioning\": 15, \"GK Diving\": 9, \"GK Handling\": 12, \"GK Kicking\": 8, \"GK Reflexes\": 14, \"Preferred Foot_Left\": 0, \"Preferred Foot_Right\": 1, \"Player Work Rate_High / High\": 0, \"Player Work Rate_High / Low\": 0, \"Player Work Rate_High / Medium\": 1, \"Player Work Rate_Low / High\": 0, \"Player Work Rate_Low / Low\": 0, \"Player Work Rate_Low / Medium\": 0, \"Player Work Rate_Medium / High\": 0, \"Player Work Rate_Medium / Low\": 0, \"Player Work Rate_Medium / Medium\": 0, \"position_zone_ATTACKING\": 0, \"position_zone_DEFENDING\": 0, \"position_zone_GOALKEEPER\": 0, \"position_zone_MIDFIELD\": 1}]'

The second cli tool is utilscli.py and this perform model retraining, and could serve as the entry point to do more things. For example, this version doesn't change the default model_name, but you could add that as an option by forking this repo.

$ python ./utilscli.py retrain --tsize 0.2

Flask Microservice

The Flask ML Microservice can be run many ways.

Containerized Flask Microservice Locally

You can run the Flask Microservice as follows with the commmand:

$ python app.py runserver

After run flask to serve a prediction against the application, run the predict.sh.

For the bourne shell:

sh ./predict.sh

For bash:

bash ./predict.sh

For windows:

https://stackoverflow.com/questions/26522789/how-to-run-sh-on-windows-command-prompt

$ bash ./predict.sh                             
Port: 8080
{
  "prediction": {
    "value_log": 15.66,
    "value_money": "6330106.07 euros"
  }
}

Containerized Flask Microservice

Here is an example of how to build the container and run it locally, this is the contents of run_docker.sh

#!/usr/bin/env bash

# Build image
docker build --tag=alonmar/mlops-value_fifa_players . 

# List docker images
docker image ls

# Run flask app
docker run -p 127.0.0.1:8080:8080 alonmar/mlops-value_fifa_players

Github Actions

When you push on main Github detect this main.yml and automatically build Container via Github Actions

Scrap the data

• Step 1: Clone the repo

$ git clone https://github.com/alonmar/Fifa-players-predict-value.git

• Step 2: Install requirements.txt

$ pip install -r requirements.txt

• Step 3: move to dir webscrapi_fifa_players

$ cd webscrapi_fifa_players

• Step 4: Start the crawler

$ scrapy crawl skills

And wait until webscraping finishes 🎉