The Simple Boxplot Method for an Effective Prediction

Abstract

This study aims to do a prediction of demand goods at a factory for 1 day ahead using double moving average method and comparing the forecasting results. Data source come from two different types of data which are complete data and clean data. Clean data was an optimal data that has been cleaned from outlier using boxplot method. The data source used in the calculation is simulation data for 945 days. Based on the test results, Shows the results of forecasting using complete data that is equal to 4692 with MAPE 6.88 while the results of forecasting use clean data that is equal to 4876 with MAPE 3.84. From these results, it can be concluded that forecasting using clean data is more accurate than forecasting using complete data because the smaller the error rate (MAPE) produced, the better the accuracy.

Pre-processing technique of Aquilaria species from Malaysia for four different qualities

“…Visualization using joint display possibly provide structure for comparing many input data between its group [25]. Boxplot has become efficient tool in the industry standard for summarize the observation value, lowest quartile, median, highest quartile, greatest observation value and outliers in one diagram [26]- [28].…”

“…There is a red line in the middle of the box which indicate the sample median. Minimum and maximum is the range values in the sample data [28]- [30].…”

“…The MATLAB overview of elements in boxplot [29] Primarily, the shorter the whiskers, the more uniform the distribution of data. The result will be accepted if 50% of the median is in the group [28], [31]. The whiskers are calculated based on the interquartile range (IQR) as mathematically shown in ( 1)-( 3) [31].…”