Preface

In the previous post, we looked at the heart failure dataset of 299 patients, which included several lifestyle and clinical features. That post was dedicated to an exploratory data analysis while this post is geared towards building prediction models.

---

Motivation

The motivating question is—_ ‘What are the chances of survival of a heart failure patient?’. _Through this walk-through, I try to answer this question while also giving a few insights on dealing with imbalanced datasets.

The code for this project can be found on my GitHub repository.

---

Quick Recap

In the previous post, we saw that —

• Age and serum creatinine had a slightly positive correlation, while serum sodium and serum creatinine had a slightly negative correlation.
• Most of the patients who died had no co-morbidities or at the most suffered from anemia or diabetes.
• The ejection fraction seemed to be lower in deceased patients than in patients who survived.
• The creatinine phosphokinase level seemed to be higher in deceased patients than in patients who survived.

(Check out the previous post to get a primer on the terms used)

#data-science #imbalanced-data #support-vector-machine #heart-disease #machine-learning #data analysis