Welcome to Team Lapiyano's submission for the Motus & University of Johannesburg Hackathon 2025.
Our project tackles the challenge of predicting whether a car lead results in a sale, using machine learning techniques with a focus on real-world impact and data-driven insights.
Predict the likelihood of a lead converting into a sale based on attributes provided in the Motus dataset. The solution aims to assist dealerships in optimizing their sales pipeline and focusing on high-potential leads.
| File / Folder | Description |
|---|---|
TeamLapiyano.ipynb |
Core notebook with full preprocessing, modeling, and evaluation pipeline |
TeamLapiyano_commented.ipynb |
Same notebook with detailed inline comments explaining every step |
Final Evaluation Standings/Results.pdf |
📊 Final performance evaluation provided by organizers |
- Model Used: XGBoost Classifier
- Preprocessing:
- Handling missing values with
SimpleImputer - Encoding categorical features
- Feature selection based on domain insights
- Handling missing values with
- Balancing Strategy:
- Applied SMOTE (Synthetic Minority Oversampling) for class imbalance
- Evaluation Metrics:
- Precision-Recall AUC
- Confusion Matrix
- ROC Curve
To run this notebook:
pip install -r requirements.txtMinimum packages:
xgboostpandasnumpyscikit-learnimblearnmatplotlibseaborn(optional)
You can find the official evaluation results in:
📁 Final Evaluation Standings/
└── 📄 Results.pdf
This file contains performance standings as assessed by the hackathon organizers.
- Team Lapiyano – University of Johannesburg Hackathon 2025
For any questions or feedback, feel free to reach out to the team or open an issue.