Прогнозирование вероятности сердечно-сосудистых заболеваний у пациентов по 12 признакам.
Prediction CVD in patients. This is my small pet-project about cardiovascular disease (CVD) in patients of different ages using 12 features. “Cardio” is the target feature.
Preliminarily, I cleaned the data and got rid of outliers. Then I made some visualizations and tested a few hypotheses. Further I compared 3 classifiers: RandomForest, XGBoost and LightGBM without tuning. XGBoost showed the best result. In the end, I made visualization with showing the most important features for the each of above classifiers. Also, there is a Power BI dashboard.
Click on "Open in Colab" below to see DETAILED conclusions and code:
PowerBI dash with some conclusions based on data:
Your feedback is welcome to me. 🙌