Data Smoothing of a Real-life Driving Test Result

A result of a real-life driving test observation usually consists of data that contains high irregularity. As a result, the shape of data trends from observations becomes difficult to know. Therefore, data that contain high irregularity in time series analysis are processed using the data smoothing method. In this study, a dataset of real-life driving tests observation of an electric vehicle is processed using the Simple Moving Average (SMA), Single Exponential Smoothing (SES), and Double Exponential Smoothing (DES) methods. Furthermore, the evaluation of each data smoothing results using Mean Arctangent Absolute Percentage Error (MAAPE). The results of data smoothing show that the lowest MAAPE is 0.252, which is generated through the SES method with data smoothing constant of 0.9. This indicates that SES with data smoothing constant of 0.9 produces data smoothing that is closest to the original data or has the smallest deviation with the original data, but data trends are difficult to know because they still contain high irregularity. Conversely, the data trend is clearly visible by the results of SMA data smoothing with a period constant of 250 with a MAAPE value of 0.793, despite having the greatest deviation with the original data. For the purpose of knowing trends in this study, SMA with a period constant of 250 shows the best results. For related studies, this trend from data smoothing could be used as a standard cycle (driving cycle in certain environments) on components and electric vehicles test.