Comparative Analysis of K-NN and Naïve Bayes Methods to Predict Stock Prices

Abstract

Abstract — Buying and selling shares is a transaction that is widely carried out at this time, especially buying and selling stocks online which are widely available in the market, to make buying and selling shares require ability or knowledge so that the buying and selling of shares are profitable, to be able to help economic players predict prices. Profit shares or not purchased in the future, this research will conduct stock price predictions using classification methods, namely K-Nearest Neighbor and Naive Bayes, to predict the stock price data used for one month in minute levels totalling 39065 data, based on prediction results. The highest results obtained were using Naive Bayes with an accuracy value of 69.38 then the K-Nearest Neighbor method with a K = 5 value of 67.25%, based on these results it can be concluded that the use of the K-Nearest Neighbor and Naive Bayes methods for prediction share price not yet owned I high accuracy, so it can be combined with other methods or by using other variable predictors.