Data Mining: The Classification Method to Predict the Types of Motorcycle Spare Parts to be Restocked

Abstract

The research intends to create an application which is able to analyse sales data in a motorcycle company to predict the types of spare parts which should be stocked. This prediction is crucial since problems are often encountered while restocking. For instance, when there have been some imprecisions occurring in deciding regarding the types of spare parts to restock, the spare parts accumulate. It can cause inefficiency in terms of storage, the products quality deteriorates due to having been stored for too long, and sometimes the best-selling products are not available in the warehouse. This application is developed with Naïve Bayes Classifier (NBC) method which has a high accuracy in predicting future occurrences. This method works by calculating the probability value in each attribute class and determining the optimal probability value. From the test results, 4500 training data with 200 sample test data has 90% similarity with the results of the restock decision without application. For 500 test data, the similarity was 96%. It is proven that this method has a high accuracy so that it can help the decision makers solved the company problem in predicting the types of motorcycle parts to be restocked.