This is an Diabetic Analysis project as part of my Internship Training at Edunet Foundation.
In response to the pressing need for improved diabetes management, this project seeks to develop models that accurately predict the likelihood of diabetes in patients. The project aims to enable proactive interventions and personalized treatment plans, ultimately enhancing healthcare outcomes for individuals at risk or already diagnosed with diabetes.
This study focuses on utilizing machine learning and predictive analytics to enhance diabetes management.
The project involves developing models to analyze patient data, including glucose levels, BMI, and family history, aiming for early detection of diabetes.
Machine learning algorithms will be implemented to stratify diabetic patients based on their risks of complications, allowing for prioritized interventions.
- LogisticRegression:
- Accuracy: 76%
- Inference: This model stands out with the highest accuracy among all the tested models. - K-Nearest Neighbors (KNN):
- Accuracy: 69.4%
- Inference: KNN gave the lowest accuracy. Further investigation or tuning might be needed to improve its performance. - Random Forest and Decision Tree:
<
- Inference: Random Forest and Decision Tree show potential but may benefit from tuning. Consider adjusting hyper parameters for better results. - Additional Metrics: - Additional Results can be obtained from the graph