Abstract
Smoothing is an exploratory data analysis approach that focuses on removing noise or unstructured pattern from data series. This study mainly aims to compare the performance of 4253HT smoother in three types of Hannings and its application in forecasting. A sinusoidal signal was used where five different levels of contaminated normal noise were applied. Overall, 4253HT smoother with Shitan and Vazifean’s Hanning performs excellently over different percentages of noise, good at preserving edges, and able to travel closely with the signal of original pattern. The smoothed rainfall data gives a lower value of RMSE than the raw data which is 12.85 and 24.25 respectively. This concludes that the trend line obtained using smoothed data is more appropriate and reliable for forecasting. These results will be useful in predicting any time series data.