This project used ensemble machine learning algorithms to predict customer subscriptions to financial products offered by a Bank. The goal was to conduct a comprehensive analysis from end to end, extracting valuable insights into customer behaviour and preferences.
The dataset used in this study contained information gathered from a marketing campaign conducted by the bank within a period spanning two years. It comprised 40,000 instances with 20 input features relating to customer demographics, engagement, and marketing outcomes.
The project was executed in various stages, including:
- Preprocessing and data cleaning.
- Exploratory analysis.
- Feature engineering.
- Implementation and evaluation of machine learning models such as decision trees, logistic regression, random forest, and XGBoost.
The Azure Machine Learning Designer was also used to create and deploy some of the models. Key steps included:
- Pipeline development and implementation of models like Neural Networks and SVM.
- Model evaluation using key metrics; the Neural Network showed superior performance.
- Configuration of an inference pipeline and deployment of the model as a web service for real-time predictions.
