Pemodelan Prediktif Harga Saham Pada Bank ABC Menggunakan K-Nearest Neighbors Dengan Pendekatan Tren

Abstract

Investing in shares in companies today often experiences several problems such as liquidity, market influence, high price volatility, and a decline in the company's ability to earn profits. One company that experiences share price fluctuations based on several factors is Bank ABC. Based on the current rapid fluctuations in Bank ABC's share prices, shareholders are expected to be able to predict when they should sell or retain their shares. The prediction technique in this research uses the K-Nearest Neighbors (KNN) algorithm in modeling stock price trend predictions using the Python programming language. By using the K-Nearest Neighbors algorithm, predictive modeling and analysis of bank ABC share price trends will be produced with the highest level of accuracy. The dataset used in this research is 1294 and has 7 features. Based on Pearson correlation, four features are obtained, namely open, high, low and price as target features that have a strong relationship in data analysis. The split data comparison used was 80:20, and using the parameter value k=3, prediction results were obtained with evaluation results with a MAPE value of 3.40% and RMSE 0.063. If the MAPE results are converted into accuracy, the results of this research obtain an accuracy of 96.60%, indicating that the prediction model used in this research is able to provide very accurate predictions.