It was one of the Project given by University under the subject called "Design Enginnering". In this subject we were supposed to develop some kind of web service & smartphone app. We were already working on Machine Learning basics and came with an idea to prepare those models and host them with REST api on Heroku and access them through Android client app.
In this project we dealt with,
- Heart Disease prediction
- Breast Cancer Detection
So we prepared,
i) FLASK RestFul API delivering the results of models.
both API:
1) http://intense-citadel-71493.herokuapp.com/heart/predict/
2) http://intense-citadel-71493.herokuapp.com/cancer/predict/
Dash separated inputs are to be provided further in above URLS.
For example, ./<any_option>/predict/1-2-3-4-5-6-7-8-9-10-11-12-13
ii) Android Client Application
In GUI we used Cardbased Views,
For communication with Web Service, We used Volley Lib
iii) Datasets
Both datasets from UCI Repository, and belong to the respective research groups.
iv) Models
Built with Sci-kit Learn, Keras, Pandas, Numpy.
Model persistance: Joblib.
v) Asynchronous Web App
AJAX Based Fully Front-end App which essentially calls Flask REST API for its operation.
It generates dynamic HTML Results with the help of Vanilla JS.