This study aims to map and model the determinants of food security. Mapping is done by cluster and biplot analysis, while modeling is done by non-linear path analysis. This research is mix-method research that combines quantitative and qualitative research. In the qualitative method, this study applies a qualitative Discourse Network Analysis (DNA) approach. Sources of DNA data come from various information in cyberspace (mass media, journals, articles, etc.) that are in accordance with the research context. In DNA data processing, statements, actors, concepts/issues, sentiments, along with the origin of the organization will be generated. As for the quantitative method, this study uses descriptive statistical analysis, biplot, cluster, and non-linear path analysis (square and cubic). The coefficient of determination for both quadratic and cubic path analysis is 0.88, which means that the influence of the independent variable simultaneously on the Y variable is 0.88, which is very strong. Thus, the model formed is quite good because the predictor variable is able to explain food security by 88% while the rest is explained by other factors outside the model. The originality of this research is the reconstruction of non-linear path analysis which is more flexible (no need for assumptions of normality and homogeneity) and is equipped with a measurement model.
Sparsity-Promoting Sensor Selection for Non-Linear Measurement Models