INTEGRATION OF NON-HIERARCHY CLUSTER ANALYSIS WITH SEMIPARAMETRIC TRUNCATED SPLINE MULTIRESPONSE REGRESSION

Cluster analysis is an approach used to find similarities in data and place them in groups. Cluster analysis is divided into two methods, namely hierarchical and non-hierarchical methods. The non-hierarchical method of several methods, namely K-Means, K-Harmonic Means, and K-Medoids, used in this study is K-Means. Then, the integration of cluster analysis with semiparametric truncated spline multiresponse regression will be carried out. This study aimed to identify and develop non-hierarchical cluster integration and semiparametric multiresponse regression using the Truncated Spline approach in the case of Farmer Inclusiveness data in West Nusa Tenggara. Modeling factors affecting the inclusiveness of farmers in NTB was conducted using cluster analysis modeling with multiresponse regression which was developed by making integration of non-hierarchical cluster analysis with semiparametric truncated spline multiresponse regression to help give Comprehensive information about farmer inclusiveness in NTB. The sample used in this study was 100 farmer groups in NTB. The results in this study produce three clusters, which have low, medium, and high inclusiveness categories. Cluster 1 is a farmer with a low inclusiveness category consisting of 42 farmer groups, cluster 2 is a farmer with a medium inclusiveness category consisting of 21 farmer groups, and Cluster 3 is a farmer with a high inclusivity category consisting of 37 farmer groups. The cluster with the highest R-squared on modeling integration of non-hierarchical cluster analysis with semiparametric truncated spline multiresponse regression is cluster 1 with R² value of 0.205. Thus, modeling using cluster integration with semiparametric truncated spline multiresponse regression shows a change in the nature of the data that has a variable relationship pattern and can help the government and farmers to determine the level of inclusiveness of farmers and examine the factors that influence farmer inclusiveness. This research’s originality is combining cluster analysis with semiparametric truncated spline multiresponse regression with integration between the two methods.