FORECASTING NATURAL GAS PRICE USING MACHINE LEARNING









Abstract

In this paper, we will be predicting the price of natural gas in US markets using statistical modelling methods. We will be studying the usage and comparison of the linear regression models: Random forest and Decisiontree model. We have used the dataset of natural gas prices in the United States in US dollars per Btu during the period of 1998 through 2020 with a total of 5800 records. Being a supervised learning algorithm Random forest performs regression using ensemble learning algorithms. By using the existing prices and comparing them, Random forest will forecast the price which will be helpful for the user to estimate the risk in bidding prices. The values of the result are marked up to 97% accuracy. This has set the path for the new opportunities these technologies could offer.


Modules


Algorithms


Software And Hardware