🍩 Diabetes Prediction with EDA and Naive Bayes

This project is designed to predict whether a patient has diabetes based on various health attributes using machine learning. The dataset contains medical data from patients, including both dependent and independent variables that help determine diabetes risk.

---

📋 Dataset Overview

The dataset consists of 768 observations and 8 numerical independent variables that are used to predict the presence or absence of diabetes (dependent variable). The target variable Outcome takes the value of 1 for a positive diabetes diagnosis and 0 for a negative diagnosis.

Columns

Feature	Description
Pregnancies	Number of times the person has been pregnant.
Glucose	Plasma glucose concentration during an oral glucose tolerance test.
BloodPressure	Diastolic blood pressure (mm Hg).
SkinThickness	Triceps skin fold thickness (mm).
Insulin	2-Hour serum insulin (mu U/ml).
BMI	Body mass index (weight in kg / height in m²).
DiabetesPedigreeFunction	A function that scores the genetic predisposition to diabetes.
Age	Age of the patient.
Outcome	Target variable indicating diabetes status (1 for positive, 0 for negative).

---

📊 Exploratory Data Analysis (EDA)

1. Data Preprocessing:

   • The data was cleaned by handling any missing values and ensuring no discrepancies in the data.

2. Visualizations:

   • Generated multiple plots to visualize the data and relationships between features:
     • Histograms to visualize the distribution of each feature.
     • Box plots to check for outliers.
     • Correlation matrix heatmap to visualize the relationships between the features.
     • 3D plot for a deeper understanding of how different features interact in relation to the Outcome variable.

---

🧠 Modeling with Naive Bayes

1. Initial Accuracy:

   • The model was first trained using Naive Bayes and yielded an accuracy of 78%.

2. Model Improvement with Cross-Validation:

   • The model was further optimized using K-Fold Cross-Validation, which improved the accuracy to 80%.

---

🎯 Results

• Naive Bayes Accuracy: 78%
• After K-Fold Cross-Validation: 80%

While the Naive Bayes model provides a good baseline accuracy of 78%, utilizing K-fold cross-validation improves the model's robustness and accuracy, providing more reliable results.

---

🛠️ Tools and Techniques

• Libraries Used: pandas, NumPy, Matplotlib, Seaborn, sklearn
• Model:
  • Naive Bayes Classifier: A probabilistic classifier based on Bayes' theorem, suitable for predicting binary outcomes like diabetes status.
  • K-Fold Cross-Validation: A technique for improving the reliability of the model by splitting the dataset into multiple training and testing sets.

---

✨ Key Takeaways

• EDA plays a crucial role in understanding the dataset's distribution and relationships between variables, allowing for better model optimization.
• Naive Bayes offers a simple yet effective approach for binary classification tasks, especially when working with normally distributed data.
• K-Fold Cross-Validation is essential for improving model accuracy and ensuring that the model generalizes well to unseen data.