Simple App which can detect Weather you are diagnoised with Diabetic or not depending up on users data provided to the application.
- Created an app that detects wheather they have Diabetics or not to help doctors with 89% accuracy .
- Data collected from Open source websites from Internet .
- Processed features to make data look's like perfect and to get good accuracy with less loss
- I had used Ada boost Classifier ,XGBoost ,Logistic ,support vector to reach best model
- Deployed model on Heroku .
- Python Version : 3.7
- Packages: pandas, numpy, sklearn, matplotlib, seaborn, selenium, flask, json, pickle
- For Web Framework Requirements: pip install -r requirements.txt
To know cor-relation between every feature i had used corr()
- To check Outliers i had used Box plot to know weather outliers present or not .And this is one of the best way to check outliers.
- To remove outliers Z-score is one of the best way to remove outliers .
First, I transformed the categorical variables into dummy variables. I also split the data into train and tests sets with a test size of 20%. I tried three different models and evaluated them using Classification Metrics. I chose Confusion Matrix Because it's better to understand how many features are going to support and not going to support . I tried Five different models:
- Support Vector Classifier: It classifies data perfectly
- Logistic Regression
- K-Nearest Neighbour Classifier
- Naive Bayes Classifier
- XGBoost Classifier
- To measure the performance of every model i had used classification metrics ,it is one of the best way to know which model is best depending up on all the metrics.
- In this step , I had deployed Model on heroku with Flask api.
- The API endpoint takes in a request with a values by end user and returns weather they have Diabetes or not . Here is URL to predict Diabetes Identification Web App