Diabetes Prediction Project

Project Overview

This project is dedicated to predicting diabetes using supervised machine learning models. We use a dataset containing medical predictor variables and one target variable, Outcome, which includes features like number of pregnancies, BMI, insulin level, and age.

Libraries Used

• Pandas: For data manipulation and analysis.
• NumPy: For numerical operations.
• Seaborn & Matplotlib: For data visualization.
• Scipy: For scientific computing.
• Scikit-Learn: For machine learning and predictive data analysis.

Data Loading and Preprocessing

• Source: The dataset is loaded from a CSV file.
• Inspection: Basic data inspection for missing values and data types.
• Details: The dataset contains 768 entries and 9 columns with no missing values.

Exploratory Data Analysis

• Box plots for visualizing data distribution.
• Z-score calculations for standardization.
• Identification and imputation of zero values in specific variables.
• Correlation analysis with heatmap visualization.
• Distribution analysis using histograms.
• Normality testing with Shapiro-Wilk test.
• Outcome-based group comparisons.

Feature Engineering

• Imputation for variables like Glucose, Blood Pressure, Skin Thickness, Insulin, and BMI.
• Detailed correlation analysis to understand feature-target relationships.

Data Visualization

• Utilization of box plots, histograms, scatter plots, and pair plots for comprehensive data visualization.

Model Preparation

• Data scaling using StandardScaler.
• Splitting data into training and testing sets.

Model Training and Evaluation

• Models Used: Logistic Regression and Random Forest Classifier.
• Evaluation Metrics: Accuracy, Precision, Recall, F1 Score, and ROC AUC Score.

Hyperparameter Tuning

• Conducting Grid Search for optimal parameters in Logistic Regression and Random Forest models.

Cross-Validation

• Evaluating model performance using cross-validation techniques.

Feature Importance Analysis

• Analyzing the importance of different features in the Random Forest model.

Conclusion

• Final recommendations based on model performance considering various evaluation metrics.

Repository Contents

• Detailed code, data, and findings are available in the Jupyter Notebook files within this repository.