/covid-ml

Machine Learning Models for the prediction of possible COVID infections using symptoms

Primary LanguageJupyter Notebook

ML model for prediction of COVID-19 cases

Machine learning models are built to predict positive COVID-19 cases from patient symptoms. The performance of the models on a relatively smaller dataset demonstrated merit in building a mobile application that could aid in rapid diagnoses and tracking of positive COVID-19 cases.

Data

A CSV file retrieved from Kaggle is used to train the model. Features:

  • age
  • fever
  • cough
  • runny nose
  • pneumonia
  • lung infection
  • travel history
  • test results

Training

Models

  • Logistic Regression
  • Random Forest
  • K Nearest Neighbor
  • Decision Tree
  • Support Vector Classifier

Results

  • Logistic Regression was the best-performing model with an accuracy score of 91%