Stroke Prediction Using Machine Learning

• By Mayank, Pritika and Shivam | 04/21 - 07/21 | Major Project

Dataset Used:- https://www.kaggle.com/fedesoriano/stroke-prediction-dataset

Description:

In this project we have performed five step process:

• 1. Evaluated and analyzed the dataset using seaborn, plotly and matplotlib libraries.
• 2. Compared various ML classifiers based on accuracy on our dataset.
• 3. Performed k-fold validation and Hyper Parameter Optimization.
• 4. Prepared the backend using Flask and linked with front end made using HTML & CSS.
• 5. Hosted the project on heroku.

Overview of Project

• READING DATA AND EXPLORING DATASET
• BASIC DATA ANALYSIS
• PANDAS PROFILING
• CORRELATION
• ANOMALY DETECTION
• MISSING VALUES
• ENCODING
• TRAIN TEST SPLIT
• TRAINING THE MODEL
• SCORES OF MODELS
• EVALUATION OF MODEL
• K-FOLD VALIDATION
• HYPER PARAMETER OPTIMIZATION
• PREPARATION OF USER GUI
• INTEGRATION WITH GUI
• HOSTING

Different Machine Learning Algorithms compared:

• 1. Gaussian Naive Bayes
• 2. Bernoulli naïve Bayes
• 3. Logistic Regression
• 4. Support Vector Machines
• 5. Random Forest
• 6. Decision Tree classifier
• 7. K-Nearest Neighbor
• 8. Gradient Boosting
• 9. Stochastic Gradient Descent
• 10. Neural nets

Requirements:-

• click==8.0.0
• Flask==2.0.0
• gunicorn==20.1.0
• itsdangerous==2.0.0
• Jinja2==3.0.0
• joblib==1.0.1
• Markdown==3.3.4
• matplotlib==3.4.2
• matplotlib-inline==0.1.2
• numpy==1.19.5
• pandas==1.2.4
• pandas-profiling==3.0.0
• plotly==4.14.3
• plotly-geo==1.0.0
• preprocessing==0.1.13
• requests==2.25.1
• requests-oauthlib==1.3.0
• scikit-learn==0.24.2
• seaborn==0.11.1
• Werkzeug==2.0.0
• wordcloud==1.8.1
• xgboost==1.4.2

How to use:

1. Dev:-

• 1. Check the requirements
• 2. Clone the repository
• 3. Open the project directory
• 4. Open and run app.py file
• 5. A web page will open in the browser, in which you can fill the details and submit
• 6. The result of prediction is shown on the next page based on input.

2. User:-

• Open "strokemsp.herokuapp.com"
• Fill in the details asked in the form
• Submit the details, and wait for a while.
• The model will show whether you are prone to stroke or not.

Results - Coming soon