<![CDATA[The release lag in official unemployment rate (TPT) data limits timely regional policy responses. This study compares nowcasting methods—U-MIDAS, Elastic Net, and Random Forest—to estimate TPT in Lampung Province by integrating monthly macroeconomic indicators (Inflation, Exports) and Google Trends Big Data (37 keywords reduced via PCA into GTREND1 and GTREND2). Pseudo-real-time evaluation with a naive baseline shows Random Forest achieves the lowest MAPE (4.89%), but U-MIDAS is recommended as the primary method due to more transparent economic interpretability (MAPE 8.00%). U-MIDAS projects TPT for August 2026 at 3.93% with a 95% prediction interval of [3.36% – 4.50%], relatively stable compared to February 2026 (3.95%). Integrating mixed-frequency indicators within U-MIDAS yields reliable estimates for regional employment policymaking, despite sample size limitations..]]>
Copyrights © 2026
