PENDEKATAN DATA TIME SERIES MENGGUNAKAN K-NEAREST NEIGHBOR BERBASIS PREPROCESING PADA PREDIKSI HARGA SAHAM

Abstract

Forex Market is a type of trade or transactions of a country's currency to another country's currency (the currency pair / pair), which involves major money markets in the world for 24 hours continuously. A trader is required to have the ability of technical analysis and fundamental good to be able to reap huge profits. Analysis traders used to predict the price in the market will rise or fall based lower threshold price (support) and upper threshold price (resistance). This study used the method of K-Nearest Neighborand a Fibonacci retracement to predict support and resistance levels. The data used in this study downloaded from the server Forex Alpari UK which consists of open, high, low, close, and volume as input data. In this study, training data and testing using test data with different time intervals. This test resulted in an accuracy rate is 50.76% on the test using the training data and test data 2 years 1 month 4 hour period of real data.