• The project goal is to build a fraud predictive model with accuracy above 90%.
• This is a binary classification task as our target represent binary data.
• I will clean data and perform feature engineering.
• I’m going to use Scikit-learn classification models and hyperparameters toning process.
• Models used for this project are: Logistic regression, Decision Tree Classifier, Gradient Boosting Classifier, Random Forest Classifier.
• In parallel I will perform search of the best hyperparameters using Scikit-learn Grid Search and Randomized Search CV.
• In my last step, I will use Scikit-learn Voting Classifier that will be trained on top 3 best performing models.

Link to full project below:

GitHub Repository for this Project