IMPLEMENTATION OF RESPONSE-BASED UNIT SEGMENTATION IN PARTIAL LEAST SQUARE (REBUS-PLS) FOR ANALYSIS AND REGIONAL GROUPING

Abstract

Housing environmental health is a key indicator of community quality of life. In West Kalimantan Province, variations in geographical and socioeconomic conditions contribute to disparities in housing conditions. This study analyzes and classifies regions based on factors influencing housing environmental health using the Response-Based Unit Segmentation in Partial Least Squares (REBUS-PLS) method. REBUS-PLS helps detect unobserved heterogeneity by identifying subgroups with different structural relationships. The exogenous latent variables include household economics, education, and housing facilities, while the endogenous variable is housing environmental health, measured through 15 indicators. The results of the SEM-PLS analysis obtained 3 paths that had a significant effect: household economics on housing facilities, household economics on education, and housing facilities on the health of the Housing environment. SEM-PLS assumes homogeneity across data, meaning all observations follow the same structural pattern. However, this assumption may not hold, especially with data representing diverse regions. To address potential heterogeneity, REBUS-PLS was applied. The analysis revealed two distinct segments, each with stronger explanatory power than the global model, as indicated by higher R² values (Segment 1 = 95.6%, Segment 2 = 91.4%, compared to 87.7% in the global model). Segment 1 consists of Landak, Sanggau, Sekadau, Kayong Utara, and Singkawang City. Segment 2 includes Bengkayang, Melawi, Ketapang, Kapuas Hulu, Sanggau, Sekadau, Sintang, and Pontianak City. These findings confirm the presence of structural heterogeneity and demonstrate that REBUS-PLS provides a more accurate understanding of the factors affecting housing environmental health across regions.