Gender Prediction Model

Project Overview

The Gender Prediction Model project aims to develop a machine learning model that can accurately predict the gender of a client using their transaction data. The project includes data preparation and analysis, model tuning, and evaluation metrics.

Getting Started

To start working on this project, follow these steps:

1. Clone the repository:

   git clone git@github.com:Melodiz/transaction-gender-prediction.git

2. Navigate to the project directory:

   cd Gender_transaction_base

3. Install the required dependencies:

   pip install -r requirements.txt

4. Download the data from Kaggle
5. Leave the unpacked data in a folder named data in the root of the repository.

Project Structure

The project's directory structure is as follows:

Gender_transaction_base/
├── LICENSE
├── README.md
├── gender_by_transaction.ipynb
├── requirements.txt
└── data/
    ├── train.csv
    ├── test.csv
    ├── mcc_codes.csv
    ├── transactions.csv
    ├── trans_types.csv
    └── test_sample_submission.csv

Dependencies

• numpy
• pandas
• matplotlib
• seaborn
• catboost
• scikit-learn
• xgboost
• lightgbm
• nltk
• gensim
• @jupyter-widgets/base
• jquery
• lodash
• plotly.js-dist-min

Project Result

The best score with k-fold cross-validation is 0.845 (ROC AUC).