Analyzing Gold Price Movements: A Comprehensive Supervised Learning Perspective

The given study explores the use of machine learning algorithms in the commodity trading scenario, specifically for gold trading. In theory, such tasks need to be accomplished with the help regression models. Taking it into the practice, implementation of various algorithms, along with testing and comparison among the models has been indicated. Linear, Multiple KNN, Decision Tree, Random Forest and Support Vector Regressions and the feed-forward neural network Multi-Layer Perceptron Regression were applied on the 3 datasets of gold price. The datasets are set in the time period and specify day-wise, week-wise and month-wise time intervals. In this work, it is inferred that neural network-based model MLP is outperforming the other basic and conventional models in terms of MAPE, MSE, RMSE and R2 score.