https://github.com/armandgiraud/betclick

https://github.com/armandgiraud/betclick

Last synced: 6 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/armandgiraud/betclick
• Owner: ArmandGiraud
• Created: 2019-10-16T17:05:04.000Z (about 5 years ago)
• Default Branch: master
• Last Pushed: 2019-10-24T19:02:55.000Z (about 5 years ago)
• Last Synced: 2024-01-17T20:34:12.704Z (10 months ago)
• Language: Jupyter Notebook
• Size: 434 KB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: readme.md

Awesome Lists containing this project

README

        
# Betclick Churn Analysis

This repo contains an Assignment for betclick

### Install:

 - clone the repo

  - `pip install -r "requirements.txt"``

 - `cd ./src`

---

Run with docker:

change env variable `PASSWORD` in docker-compose

`sudo docker-compose up --build`

### Fit

- fit the model on data using:

    ```python main.py```

The script will:

    - download the data, and optionnally ask a password for unzipping

    - label the data and drop leaky rows

    - preprocess and write serializables necessary for inference on disk

### Predict

`python main.py --predict`

- predict whether each customer in a subsample of the dataset is a potential churner.

- write a file on disk in the preds folder

the first column rerpresents customer_key

the second column is the target 

| customer_key | is_churner |

|--------------|------------|

| 10390929     | True       |

| 10390926     | True       |

| 10390926     | False      |

| 10390926     | True       |

| 10390926     | False      |

| 10390926     | False      |

#### Predict on private dataset

`python main.py --predict --private_file "my_data_file.csv"`

The data should have the same format as the original one, it might be necessary to handle mix typed columns.