Forecasting <scp>COVID</scp> ‐19 cases using dynamic time warping and incremental machine learning methods

The investment of time and resources for developing better strategies is key to dealing with future pandemics. In this work, we recreated the situation COVID-19 across year 2020, when pandemic started spreading worldwide. We conducted experiments predict coronavirus cases 50 countries most during 2020. compared performance state-of-the-art machine learning algorithms, such as long-short-term memory networks, against that online incremental algorithms. To find best strategy, performed test three different approaches. first approach (single-country), trained each model using data only from country were predicting. second one (multiple-country), a countries, used countries. third experiment, applied clustering calculate nine similar consider two be if differences between curve represents series are small. do so, similarity measures (TSSM) Euclidean Distance (ED) Dynamic Time Warping (DTW). TSSM return real value distance points in which can interpreted how they are. Then, models more was predicted one. ARIMA baseline our results. Results show idea very effective approach. By it ED, obtained RMSE single-country multiple-country approaches reduced by 74.21% 74.70%, respectively. And DTW, 74.89% 75.36%. main advantage methodology simple fast apply since based on data, opposed complex methodologies require deep thorough study number parameters involved spread virus their corresponding values. made code public allow other researchers explore proposed methodology.