This project focuses on classifying customers into distinct categories based on their behaviours and characteristics. Understanding and classifying customers into different segments in the modern business environment can significantly help marketing and improve customer service. This pipeline uses Random Forest and Logistic Regression to achieve this goal.
The dataset consists of various client details, like age, work experience, family size, and more. These details have been pre-processed and used to train machine learning models to classify the clients.
- Random Forest Classifier: Achieved an accuracy of around 53% on unseen data.
- Logistic Regression: Provided a similar accuracy, with the added advantage of offering class probabilities.
These scores are consistent with top models on Kaggle for this dataset.
- Due to time constraints, there was a rapid exploration of feature engineering possibilities.
- While the dataset is detailed, it has a limited sample size.
- Additional information about the clients might have further improved the model's performance.
This project serves as an initial exploration into client classification using machine learning, but there's ample potential for subsequent work:
- Experimentation with more sophisticated models or ensemble techniques.
- Consideration of deep learning methods.
- The potential collection of more data or integration with external data sources.
- Further refinement of features and possibly integration of domain-specific insights.
Although this project is primarily a personal endeavor and learning experience, insights or suggestions are always appreciated. Feel free to open an issue or submit a pull request.
This project falls under the MIT License. Refer to the LICENSE file for more details.