Predicting the power of a combined cycle power plant using machine learning methods

The gas turbine is the most important part of the combined cycle power plant that generates the total electric power from the fuel to provide it to the houses, schools, and other facilities in the country. Thus, it is important to predict the power to increase and maximize profit. This paper compares four machine learning algorithms which are Multiple linear Regression, Multilayer perceptron, K-Nearest Neighbors, and Random Forest Algorithm. The dataset consists of 9,568 observations and four inputs which are ambient temperature, ambient pressure, relative humidity, and exhaust vacuum that will be used to train the prediction of the total electric power consumption which is the output. The best result was shown by using the Random Forest Algorithm with the mean absolute error of 2.3013 and root mean square error with 3.3061.