PREDICTIVE DISTRIBUTION TO DETERMINE LEARNING MODEL AT THE STRATEGIC COMPETENCE LEVEL OF STUDENTS IN STATISTICS GROUP COURSE

Abstract

The problem of this research comes from a situation or condition that is not static. The description of these problems is the condition of the learning system, which tends to change due to the Covid-19 pandemic, causing learning conditions to be dynamic. From a statistical perspective, the dynamic situation can be modeled using a predictive distribution approach, so its characteristics can be studied. The purpose is to provide policy recommendations on appropriate learning models for lecturers in improving students' strategic competence, which is an ability that students need to master in solving various mathematical problems. The main discussion of this paper consists of three parts: clustering, predictive distribution, and statistical inference. The purpose of clustering is to group students based on test results to determine the level of strategic competence. In addition, clustering is also used as an initial process to predict students' strategic competence level if the learning used is still the same. The benefits of statistical inference in the distribution procedure in this study are used to determine the type of data distribution from each arrival of new information or data. The results of the statistical inference determine whether or not it is necessary to update the learning model of the lecturer. This research produce a new alternative statistical inference needed to make decisions. Based on the simulation results and discussion, the use of a predictive distribution approach to predict dynamic data is very appropriate. Distribution approach can use for detecting changes in new data distribution with historical data for the dynamic condition. If the changes are insignificant, direct instruction can still be used for the learning model in statistics course. A new learning model is recommended for the statistics group course at a higher level when the changes are significant.