Cardiovascular_Risk_prediction Problem Statement: The dataset is from an ongoing cardiovascular study on residents of the town of Framingham, Massachusetts. The classification goal is to predict whether the patient has a 10-year risk of future coronary heart disease (CHD). The dataset provides the patients’ information. It includes over 4,000 records and 15 attributes.
Approach: I started the study by handling missing values, our dataset was having many columns with null entities I have removed null values using KNN Imputer and Simple imputer I've seen how smoking, systolic BP, diastolic BP, BMI, Heart rate, glucose, hypertensive, cholesterol, diabetes, etc affects the person.
Factors like Blood Pressure, Glucose Level, Age had created a huge impact on a person's heart condition. We checked the correlations between the factors. Handled the class Imbalance using SMOTE and experimented with a combination of SMOTE + Tomek links. SMOTE gave a good result of 50-50 class balanced data.
Then I started building classification models. I started with Logistic regression with default parameters but I did not get a good score. Then I used a Support Vector Machine with various Kernel tricks with respective hyperparameters. I have used linear kernel, polynomial kernel and gaussian radial based kernel. Polynomial and rbf kernel gave good scores though it took a large amount of time in rendering.
Then I tried a neural network with a single hidden layer there we got decent numbers but less as compared to the rbf poly kernel. Then I tried the Random Forest Classifier and I got great Scores. Random Forest Classifier performed well. And at last I've tried XGBoost Classifier and it had performed really well and I got the best scores.
Conclusion: XGBoost Classifier as compared to other Models, so i conclude XGBoost is my optimal model for use and we can use this model for further in predicting Cardiovascular risk.