Prediction of COVID-19 Possibilities using KNN Classification Algorithm

Abstract

This paper studies the different machine learning classification algorithms to predict the COVID-19 recovered and deceased cases. The k-fold cross-validation resampling technique is used to validate the prediction model. The prediction scores of each algorithm are evaluated with performance metrics such as prediction accuracy, precision, recall, mean square error, confusion matrix, and kappa score. For the given dataset, the k-nearest neighbour (KNN) classification algorithm produces 80.4 % of predication accuracy and 1.5 to 3.3 % of improved accuracy over other algorithms. The KNN algorithm predicts 92 % (true positive rate) of the deceased cases correctly with 0.077 % of misclassification. Further, the KNN algorithm produces the lowest error rate as 0.19 on the prediction of accurate COVID-19 cases than the other algorithm. Also, it produces the receiver operator characteristic curve with the output value of 82 %.

A Review on Machine Learning Approaches in COVID-19 Pandemic Prediction and Forecasting

“…In epidemiology, logistic regression is commonly used in the time series regression fitting to predict the likelihood of the occurrence of a certain disease due to its simplicity and efficient calculation (3,20). It uses the sigmoid function to perform predictive analysis based on the relationship between 0/1 or binary dependent variables (25). The logistic growth of COVID-19 is characterised as in Figure 2A, in which the spread starts with a slow increase in growth, then grows fast near the peak of the incidence curve, and latterly a slow growth phase near the end of the outbreak (3).…”

“…Despite the involvement of many excellent models, Chakraborty et al (33) stressed that predicting and forecasting COVID-19 is challenging primarily due to seven major factors, including limited availability of data and extreme sources of uncertainty resulting in no gold standard for accurately forecasting the pandemic data. DT observes an object's features and trains a model which is represented in the form of a binary tree to predict data in the future (25,28,29). Figure 3A shows that the prediction is made by taking the root node of the binary tree with a single input variable, splitting the dataset based on the variable, and its leaf nodes have resulted as the output variable (25,26).…”

“…DT observes an object's features and trains a model which is represented in the form of a binary tree to predict data in the future (25,28,29). Figure 3A shows that the prediction is made by taking the root node of the binary tree with a single input variable, splitting the dataset based on the variable, and its leaf nodes have resulted as the output variable (25,26). The Gini index function is often used to determine the split or impurity level for the predictions.…”

“…SVM is a type of supervised ML algorithm that works well on both classification and regression (19,22,25). It separates the variables or so-called support vectors using a hyperplane by maximising the margin between two classes, obtained from the distance of decision boundaries as illustrated in Figure 1 (13,25,26).…”

“…SVM is a type of supervised ML algorithm that works well on both classification and regression (19,22,25). It separates the variables or so-called support vectors using a hyperplane by maximising the margin between two classes, obtained from the distance of decision boundaries as illustrated in Figure 1 (13,25,26). The main advantages of SVM lie in its capability to recognise the predictor non-linear pattern and can improve the forecasting cases based on the history (13,27).…”