Coffee Price Pridiction Using Machine-Learning Techniques:

Abstract

Commodity market decision making is a very challenging task because it is difficult to make a good decision in a timely manner even if the decision is made in time may not found to be the right one, it is significant to course correct and moves on. Forecast about commodity market with high accuracy movement produce profits for users of the market. In this study machine-learning prediction model is developed by comparing different prediction algorithms, the one with high performance to predict the un-for-seen data is selected. An approach to forecast future market track of the Ethiopian Commodity Exchange (ECX) data has been introduced based on chart patterns recognition by using machine learning Regression model. Models are built through different algorithms including Linear Regression, decision tree Regression (CART), SVR, KNN, and LSTM. Where the Results -6614722.039, 0.909470, 0.910699, -0103426 and 0.91523 are obtained respectively, compared and discussed in details in chapters. Important patterns to support decision making in commodity trading had been found out. In order to visualize the result, ARIMA visualization technique is also introduced Time series prediction is conducted and a fit span of time for the commodity market data is examined. After a series of algorithms comparison and experimental analysis, LSTM is used to predict the trade data. This paper attempted to predict the future market direction based on ECX historical data. to decide whether to buy or hold the commodity using machine-learning techniques.