Project Classification Based on Master Plan under Nation Strategy Using Text Mining

Abstract

         The objectives of this research were to compare the effectiveness of text mining technique used for developing a model to classify projects based on a master plan under a national strategy and measure the accuracy of a project classification model based on a master plan under a national strategy from the development model. The dataset was 1,150 projects collected from the National Monitoring and Evaluation System (eMenscr). In this research, the text mining technique was used to produce a project classification model based on a master plan under a national strategy which four techniques: Support Vector Machine, Random Forest, Naive Bayes, and k-Nearest Neighbors.

            The results showed that Random Forest was the most effective technique, with accuracy at 82.50%, precision at 83.30%, recall at 82.50%, and F-measure at 82.50%. Moreover, Random Forest had the highest accuracy at 90.91% in classifying 22 new projects.