Stroke Prediction
DESCRIPTION
This project dives into stroke probability for adult people, and the possibility of prediction of a stroke based on a set of characteristics/attributes.
RESULT
Results are presented in a report where logistic regression models were built. The final model achieved an accuracy of 92.5%.
The numerical variables chosen to include in our logistic regression model as predictors, were Age, BMI and Glucose Level. As can be seen above, the three attributes were found to have an effect on the probability of an adult person having a stroke. A similar process was followed to obtain the categorical predictors.
After performing backward elimination on the initial set of variables, the final model looked like this.
If you wish to read the complete report, click here.