For several decades, time-series forecasting has been an engaging research area. It is an essential domain of Machine Learning (ML) that is mainly ignored. It is necessary because prediction problems have a time feature, which makes time series problems more difficult to tackle. Forecasting of many applications such as weather, sales, ECG patterns and even COVID-19 spreads are possible with time series techniques. Inspired by these applications, many scholars have worked on effective forecasting techniques. This paper presents a comparative study of the time series models implemented on India’s real-time data of COVID-19. The study aims to estimate the mortality rate of coming 10 days by the interpretation of actual data. Two predictive algorithms, Holt’s Linear Exponential Smoothing (HLES) and Autoregressive Integrated Moving Average (ARIMA) have been applied. To accomplish the objective and check the model accuracy, two selection criterion methods, Root Mean Square Error (RMSE) and Akaike Information Criterion (AIC), have been used to calculate the lowest values. The results depict that the HLES model has generally outperformed ARIMA. Adding to this, HLES model has good accuracy in forecasting the mortality rate compared to ARIMA. Moreover, if we face similar circumstances again in the future, then the proposed algorithm can be used to prevent the earlier phase of the outbreak.