DEVELOPMENT A FORECAST MODEL FOR PREDICTING SCHOOL UNIFORM ORDERS USING DATA MINING TECHNIQUES

The development of a forecast model for predicting school uniform orders using data mining techniques aimed 1) to create and test a school uniform order forecast model. 2) to find the efficiency of the system and the satisfaction with the forecasting system and can predict student uniform orders using data mining techniques. The researcher used a method to categorize the results of the student uniform order data classification. The model developed with Cross Validation 10 Folds method using the Decision Tree (J48) algorithm, Naïve Bayes algorithm, and the Neighbour search algorithm found that the Decision Tree (J48) algorithm had an accuracy of 86.60%, an accuracy of 87.10%, a recall of 84.50%, and a balance of 86.40%. The Naïve Bayes athlete has an accuracy of 79.77%, an accuracy of 78.70%, a recall of 78.70% and a balance of 75.66%, the Near Neighbour Algorithm has an accuracy of 82.65%, an accuracy of 81.60%, a recall of 80.44%, and a balance of 81.88% The results showed that the classification of student uniform order data using the Decision Tree (J48) algorithm, the data classification was 84.84% accurate and there were errors. 15.16 percent, Naïve Bayes algorithm, 74.11% data classification accuracy and 25.89 percent error, and Proximity algorithm. The data classification results were 81.60 percent accurate and there was an error of 18.40 percent, respectively. It can be seen that the Decision Tree (J48) algorithm has the highest student-order data classification value than any other technique students effectively.