Inspiration
I am always curious about how machine learning was able to predict and generate future data using regression. I am passionate about coding and wanted to learn more about ml and the mathematical algorithms implemented in them.
What it does
My web application uses polynomial regression to predict COVID 19 cases in several countries using JSON data that was provided by Github. The application trains a model using the data provided and gets the most optimal values for a,b and c. These weights are then put into the ax^2 +bx +c equation with an x value to get the number of cases.
How I built it
I know running the training process in the browser would be very slow. Hence decide to use node.js and create my own API for the front end to interact with. For this, I used node.js and some other libraries to create the front end and back end. For the front end, I mainly used HTML, CSS and javascript to interact with the API. I created several posts and get requests allowing the front end to interact with. Most of the training and processing are done in the node.js allowing the front end to be fast and efficient.
Challenges I ran into
The challenges I faced while creating this web application was mainly getting an accurately predicted result. At first, I used linear regression which took me 2 days to understand the concept and how the mathematical algorithms work. However, soon I realized that linear regression was not the best fit for the COVID-19 data. Due to this I started learning polynomial regression and was able to get better results. Another section of the web application that challenged me was the front end and getting the data in a certain way. This was mainly working with the uploaded JSON as well as the date that was selected by the user.
Accomplishments that I'm proud of
I am proud to be able to implement polynomial regression and understand the mathematical algorithm used in it. I was mainly proud of being able to train all the various covid-19 countries with a 100k iteration , and seeing the weights coverage to a more optimal value.
What I learned
I learned more about the front end as well as the back end this allowed me to understand more about designing a web application. I also learned about polynomial, linear regression and how mathematically these regression work.
What's next for Best Beginner Hack-Machine Learning-COVID 19 Predictor
I would like to use the mappa.js to provide much better visualization of the COVID 19 prediction across the world.
Log in or sign up for Devpost to join the conversation.