The goal of this project is to introduce us to ML model creation processes. It's a part of the ML course at Faculty of Mathematics and Information Science at Warsaw University of Technology.
- EDA (Exploatory Data Analysis) - go into a dataset, understand features and find relationships beetwen them.
- Feature Engineering - select features and transform them to maximize a performance of out future models
- Create models - fit appropriate models to our problem and tune hiperparameters
- Create a summary report
- Validate another team
Additionally we're being validated by one team as well as validating the next team. The validators need to help the modelers and correct them in case of wrong decision making at each stage of the project. Finally they'll make a short validation report of what they've done to improve their colleagues models.
We are basing on data from the Kaggle dataset which represents a direct marketing campaign of a Portuguese banking institution. Our target is to predict if a client subscribed to a term deposit or not (variable y) by using available features collected during a meeting/call. During whole process we only have an access to a bank_train_data.csv which is 70% of the whole dataset. With the rest 30% we'll be validated.