Ordinal Logsitic Regression to Identify the Influence of Visitor Reviews Using Latent Dirichlet Allocation on Baloga Ratings

Ordinal logistic regression analysis is an analysis used to determine the effect between predictor variables and response variables that are categorical with an ordinal scale (level). This research aims to pull data based on reviews given by BALOGA visitors on Google Maps Review. The data is extracted using web scraping and Latent Dirichlet Allocation analysis is performed on the results obtained from web scraping. Before the analysis is done, data preprocessing needs to be done first to get optimal results. Latent Dirichlet Allocation results show there are four topics with a coherence value of 0.475. Latent Dirichlet Allocation produces the weight of each review on four topics. Weight data is used to perform ordinal logistic regression analysis. Ordinal logistic regression analysis was conducted to identify the influence of visitor reviews on BALOGA's rating on Google Maps Review. The results of ordinal logistic regression analysis produced 3 significant predictor variables, after going through the backward elimination stage for 1 stage, namely the weight of a review into topic 1 (X1), the weight of a review into topic 3 (X3), and the weight of a review into topic 4 (X4). The Lipsitz value shows a number greater than the chi-square value (LR = 22.581 > X2 0.05,4= 9.488), indicating that the resulting model is not suitable. This model has a classification accuracy of 67.29%.