Diabetes Prediction Web App

Overview

This project aims to create a web application that predicts the likelihood of an individual being diabetic based on their glucose level, BMI, and age. The application uses a machine learning model trained on the Random Forest algorithm.

Steps

• Data Collection: The data used for this project is taken from Kaggle.
• Data Preprocessing: The dataset is preprocessed to remove outliers and features not relevant to the prediction (Pregnancies, BloodPressure, SkinThickness, Insulin, DiabetesPedigreeFunction).
• Feature Engineering: Features are extracted and split into input features (glucose, BMI, age) and target feature (Outcome).
• Data Balancing: The dataset is balanced using the Synthetic Minority Over-sampling Technique (SMOTE) to address class imbalance.
• Normalization: Input features are normalized using the StandardScaler to ensure consistent scaling across features.
• Model Training: The RandomForestClassifier is trained on the preprocessed and normalized data to predict the likelihood of an individual being diabetic.
• Web App Development: The Flask framework is used to develop a web application that takes input from users (glucose, BMI, age), uses the trained model to make predictions, and displays the results.

Techniques Used

• Machine Learning: Random Forest Classifier
• Data Preprocessing: Outlier removal, feature selection, data balancing
• Normalization: StandardScaler
• Web Development: Flask

Prediction

predict_diabetes(glucose, bmi, age): Function that takes glucose, BMI, and age as input, normalizes the input features, and returns the prediction and probability of being diabetic.

Usage

• Clone the repository.
• Install the required libraries pip install -r requirements.txt.
• Run app.py.
• Access the web application in a browser.

Web App