COVID-19 Cases Predictor

Creator: Piero Orderique

OrderiqueStudios

This COVID-19 Cases Predictor program takes a machine learning approach to battling COVID-19 cases in America. Written in Python using tkinter, matplotlib, and scikit-learn, the easy-to-use UI permits users to display data of their choosing and run a polynomial regression model on the data to see how cases will tend towards in the future.

My Purpose

Mostly inspired by the recent spikes in the COVID-19 pandemic in the United States, I decided to use machine learning to tackle the problem starting at a fundamental community level. Most visuals out on the internet showcase either worldwide, national, or state data. While this data is beneficial to all, I believe that showing visuals at a county level will help bring a more personal awareness of how the pandemic has affected the community around us. Not only does this program make these visuals available to users, but by allowing them to run a regression model, users can further see the potential implications on their communities if the current state of the pandemic continues to grow.

Features

  • National, State, and County Selection Data
  • Regression Button that trains the model
  • Navigation Bar to zoom into more or less recent dates
  • Evaluation Summary of Model when tested with "outside" data

How it was Built

Python was used for the entire program along with tkinter, matplotlib, and scikit-learn libraries.

What I learned

How to embed graphs into tkinter windows, how to run a polynomial regression model on COVID-data, how to create training and testing data sets

Challenges

A new Navigation Bar object was created every time a new graph was selected, eventually covering the entire screen. Regression model generalization in order to use one function to handle all possible graph selection events.

Future Goals

The current goal is to further develop the regression model to where the program can distinguish between logistic, polynomial, and exponential trendlines and make a decision to which one fits the data best. Furthermore, I would love to take this project into augmented reality to showcase data in 3 dimensions.

Built With

Share this project:

Updates