Diabetes Detection Machine Learning Final Year Project with Source Code

Project Features in this Final Year Project

• Diabetes prediction using machine learning
• Healthcare data analysis
• Flask-based prediction web app
• Pima Indians Diabetes dataset support
• Feature selection
• Correlation matrix analysis
• Outlier detection
• Outlier treatment
• Feature transformation
• Feature scaling
• StandardScaler preprocessing
• Multiple model comparison
• Logistic Regression model
• Naive Bayes classifier
• K-Nearest Neighbors model
• Decision Tree Classifier
• Support Vector Classifier
• Confusion matrix evaluation
• Hyperparameter tuning
• Grid search support
• User input form for prediction
• Diabetes result prediction
• Jupyter Notebook support
• Project report content support
• Suitable for academic viva explanation

Dataset Details in this Final Year Project

Dataset domain:
Healthcare / diabetes diagnosis prediction

Dataset objective:
Predict whether a patient has diabetes based on diagnostic measurements.

Dataset background:
The dataset is originally from the National Institute of Diabetes and Digestive and Kidney Diseases. The dataset contains diagnostic measurements for female patients at least 21 years old of Pima Indian heritage.

Dataset features:

Feature	Description
Pregnancies	Number of times pregnant
Glucose	Plasma glucose concentration
BloodPressure	Diastolic blood pressure
SkinThickness	Triceps skin fold thickness
Insulin	2-hour serum insulin
BMI	Body mass index
DiabetesPedigreeFunction	Diabetes pedigree function
Age	Age in years
Outcome	Class variable, 0 or 1

Target variable:

Outcome

Prediction output:

• Diabetes not indicated
• Diabetes indicated

---

Data Preprocessing Features in this Final Year Project

1. Feature Selection

• Selects relevant features from the diabetes dataset
• Uses correlation matrix to identify feature relationships
• Helps reduce irrelevant or highly correlated features
• Improves model interpretability
• Helps reduce overfitting
• Supports better model performance

2. Correlation Matrix

• Displays relationship between dataset variables
• Helps identify strongly related diagnostic features
• Supports feature selection decisions
• Useful for project report visualization

3. Outlier Detection

The project identifies outliers using:

• Visualization
• Box plots
• Scatter plots
• Histograms
• Interquartile range method

4. Outlier Treatment

The project treats outliers by replacing extreme values using methods such as:

• Mean
• Median
• Mode

5. Feature Scaling

• Uses StandardScaler
• Standardizes feature values
• Reduces scale imbalance between features
• Improves model training performance
• Helps classification algorithms work more effectively

---

Machine Learning Features in this Final Year Project

The project compares multiple classification algorithms:

1. Logistic Regression
   Used for binary classification and selected as the main model for the web app.
2. Naive Bayes Classifier
   Probabilistic classifier based on Bayes theorem.
3. K-Nearest Neighbors
   Instance-based classification algorithm.
4. Decision Tree Classifier
   Tree-based supervised classification model.
5. Support Vector Classifier
   Classification algorithm that separates classes using an optimal hyperplane.

---

Model Accuracy Comparison in this Final Year Project

Based on the uploaded documentation, the confusion matrix accuracy values are:

Algorithm	Accuracy
Logistic Regression	0.8053435114503816
Naive Bayes Classifier	0.7709923664122137
K-Nearest Neighbors	0.7366412213740458
Decision Tree Classifier	0.7404580152671756
Support Vector Classifier	0.7977099236641222

Selected model:
Logistic Regression is selected because it shows the highest confusion matrix accuracy among the compared algorithms.

---

Hyperparameter Tuning Features in this Final Year Project

The project applies hyperparameter tuning using Grid Search.

Logistic Regression hyperparameters used:

solvers: ['newton-cg', 'lbfgs', 'liblinear']
penalty: ['l2']
c_values: [100, 10, 1.0, 0.1, 0.01]

Tuned Logistic Regression accuracy mentioned in documentation:

0.7760416666666666

---

Flask Web App Features in this Final Year Project

• Browser-based diabetes prediction form
• User enters diagnostic values
• Flask backend processes input
• ML model predicts diabetes possibility
• Displays result on web page
• Lightweight Flask application
• Suitable for local execution
• Can be extended for deployment
• Useful for project demonstration

Example input attributes:

• Glucose
• Age
• Blood Pressure
• Other diagnostic values depending on form design

• Complete Diabetes Detection source code
• Suitable for final year project, major project, and minor project
• Project report content can be prepared from included modules
• Python machine learning project
• Healthcare prediction system
• Flask web application
• Dataset preprocessing
• Feature selection
• Correlation heatmap
• Outlier detection
• Outlier treatment
• Feature scaling
• Multiple model comparison
• Logistic Regression selected model
• Confusion matrix evaluation
• Hyperparameter tuning
• User input based prediction
• Jupyter Notebook support
• Useful for viva, source code review, project report writing, and project demonstration

1. Open project folder:

   cd Diabetes-Detection

2. Optional: install virtualenv:

   pip install virtualenv

3. Create virtual environment:

   virtualenv myenv

4. Activate virtual environment on Linux/macOS:

   source myenv/bin/activate

5. Install requirements:

   pip install -r requirements.txt

6. To run notebook:

   jupyter notebook

7. To run Flask app:

   flask run

8. Or run using Python:

   python app.py

9. Alternative:

python main.py

11. Open in browser:

http://localhost:5000

This project has no login credentials because it is not described as a login-based web application.

Account Type	Username	Password
Administrator	Not required	Not required
User	Not required	Not required

Credential note:
The Flask app works through a prediction form where users enter diagnostic values and view prediction results.

---

Medical Disclaimer

This project is for educational and academic demonstration purposes only. It should not be treated as a medical diagnosis system. Diabetes diagnosis must be confirmed by qualified medical professionals using proper clinical tests and medical standards.

Copyright Notice Original Software: Copyright (c) 2023 Nilesh Parab Licensed under the MIT License. Modifications, documentation, testing, configuration, packaging and additional components: Copyright (c) 2026 FileMakr.com The original copyright notice and MIT License continue to apply to the original software. FileMakr.com claims copyright only over its independently created modifications, documentation, reports, designs and other additional materials. Recommended Website Disclosure This product contains open-source software originally developed by Nilesh Parab and released under the MIT License. The original copyright notice and complete MIT License are included in the downloadable source-code package. FileMakr.com does not claim ownership of the original open-source software. The price charged by FileMakr.com covers value-added services such as source-code verification, testing, project organization, documentation, report preparation, packaging, delivery, configuration guidance and technical support. Customers receive the original software under the terms of the MIT License. Any third-party libraries, APIs, images, fonts, datasets, templates or other external resources included in the project may be governed by their own respective licences and terms of use.