nmtruong93 / gender_classification Goto Github PK

1. Create folder named data in the root directory data folder directory should be like this

    data
    ├──trainingData.csv
    ├──trainingLabels.csv
   
   

2. Install python 3.10

3. Create a virtual environment

   python -m venv venv

   source venv/bin/activate

   python -m pip install --upgrade pip

4. Install the required packages

   pip install -r requirements.txt

5. Run the pipeline_train.py

   • Using IDE
     • Open the project in your favorite IDE
     • Run the file pipeline_train.py
   • Using command line

      cd gender_classification
        
      export PYTHONPATH="${PYTHONPATH}:${PWD}" 
        
      python gender_prediction/pipeline_train.py

   Please note that the training process will take a few minutes to complete. We use Optuna to find the best hyperparameters for the model.

6. The trained model will be saved in the tmp folder

Run the file pipeline_prediction.py

Label:

• Female: 0.0
• Male: 1.0

Overall Gender Score: 0.8257492829234192

Overall metrics:

              precision    recall  f1-score   support

         0.0       0.91      0.98      0.94     11703
         1.0       0.89      0.68      0.77      3297

    accuracy                           0.91     15000
   macro avg       0.90      0.83      0.86     15000
weighted avg       0.91      0.91      0.91     15000