APPLICATION OF SPATIAL DATA REGRESSION ON FACTORS AFFECTING OPEN UNEMPLOYMENT RATE IN INDONESIA Nabila Aulia Putri Ganessa, Sekti Kartika Dini*
Department of Statistics, Faculty of Mathematics and Natural Sciences, Universitas Islam Indonesia
*sektidini[at]uii.ac.id
Abstract
The Open Unemployment Rate (TOUR) is an indication of the working age population which is included in the unemployment group. In 2021 the number of open unemployment in Indonesia will decrease by 0.58% from 2020. The economic activities of a region can be influenced by the economic activities of other adjacent areas. In spatial analysis, eliminating an outlier can result in changes in the composition of the spatial effect on the data. So in this study, discarding outlier data is an inappropriate step. To observe the spatial effect in Indonesia requires 34 provinces to provide the right information at each point. Analyzing spatial data that has outliers, robust regression is used. In robust regression, to obtain the best guess, it is necessary to do iterative calculations. So that the estimated value is obtained which has the smallest standard error parameter. The appropriate method is the Robust M Estimator which is solved using Iteratively Reweighted Least Square (IRLS) with Tukey^s Bisquare function. The results showed that the best model of factor analysis that affects TOUR spatially is the Robust Spatial Autoregressive (RSAR) model. Shown by the coefficient of determination of 98.48% with a Mean Square Error of 0.05231617. The results showed that the best model for analyzing factors that affect TPT spatially is by using the Robust Spatial Error Model (RSEM). Shown by the coefficient of determination of 81.39% with a Mean Square Error of 0.5896. Then it is also supported by the MAPE value of 13.92%. The error value of the TPT variable (λ-) is 0.5009. This means that the TPT of each province will have an effect of 0.5009 times the average percentage of TPT for each province that is a neighbor. If GRDP, LFPR, HLS and HDI are considered constant and when PPM decreases by 1%, then TPT will decrease by 0.0904%. If PPM, TPAK, HLS, and HDI are considered constant and when GRDP increases by 1%, then TPT will decrease by 0.0274%. If PPM, GRDP, HLS and HDI are considered constant and when LFPR increases by 1%, then TPT will decrease by 0.0625%. If PPM, GRDP, TPAK and HDI are considered constant and when HLS increases by 1 year, then TPT will decrease by 0.6565%. If PPM, GRDP, TPAK and HLS are considered constant and when HDI increases by 1 unit, then TPT will decrease by 0.2153%.
Keywords: bisquare tukey, IRLS, outlier, robust regression, The Open Unemployment Rate.
