Covid-19 case modeling in Java Island using a spatial model, GSTAR(1;1), with modified spatial weights: Queen contiguity weight matrix.

The Covid-19 pandemic that has occurred in the past two years is still the main topic, especially when this pandemic ends. The essence of the Covid-19 pandemic is the movement of people (as virus carriers) from one place to another. The move allows a location to be interconnected with other locations in the increase in Covid-19 cases. This linkage is a spatial relationship between locations. The same applies to the relationship between time. The main factor is the incubation period of the virus, which causes the increase in cases today to be influenced by cases in the past few days. The two correlations between spatial and time are combined to form the basis for space-time modeling. The model used is the Generalized Space-Time Autoregressive (GSTAR) model, limited to the order of 1:1. The uniqueness of this model is the presence of a weight matrix representing the spatial correlation. In this study, the weight matrix was modified using the Queen Contiguity Weight Matrix, which is based on neighbors directly adjacent or not between locations. The data used in this study is data on positive confirmed cases of Covid-19 per district in three provinces on the island of Java, namely Banten, DKI Jakarta (the state capital), and West Java. The total locations used are 41 districts/cities. The modeling step is to calculate the weight matrix, then estimate the parameters, and finally, the prediction for the following five periods (which is also the primary goal of this study). The results obtained predictive values for 41 districts/cities on June 12-16, 2022. [ABSTRACT FROM AUTHOR]