Comparison of Medium-Term Load Forecasting Methods (Splitted Linear Regression and Artificial Neural Networks) in Electricity Systems Located in Tropical Regions Agus Setiawan Electrical Engineering Department Universitas Indonesia Jakarta, Indonesia agus.setiawan02@ui.ac.id Zainal Arifin Engineering and Technology Division PLN Jakarta, Indonesia zainal_pln@yahoo.com Budi Sudiarto Electrical Engineering Department Universitas Indonesia Jakarta, Indonesia budi@ui.ac.id Fauzan Hanif Jufri Electrical Engineering Department Universitas Indonesia Jakarta, Indonesia fauzanhj@ui.ac.id Qasthalani Haramaini Electrical Engineering Department Universitas Indonesia Jakarta, Indonesia qastil@yahoo.com Iwa Garniwa Electrical Engineering Department Universitas Indonesia Jakarta, Indonesia iwa@ui.ac.id Abstract—Load forecasting for the medium term to supply power plants within 1 month, can optimize the economic dispatch of generators that will be used to supply large systems. The accuracy of forecasting the half-hourly load will result in a more efficient electricity supply in a system that uses flat rates on the utility side. In this paper, the author tries to compare the forecasting methods of linear regression, Artificial Neural Networks, and Splitted Linear Regression. This method is applied to the largest system in Indonesia, a country located at the tropical region which has different characteristics from countries that have four seasons. At the end of this study, it can be concluded that the Splitted Linear Regression method has the highest performance with the lowest MAPE value of 2.63%. Keywords—medium-term load forecasting, Splitted linear regression, MAPE, deep learning load forecasting I. INTRODUCTION Electricity is a human need in everyday life and one of the main driving factors for a country's economy growth. In providing adequate electricity to meet electricity needs at any time reliably, it is necessary to estimate the load for planning electricity production. Because the electricity supply has an uneasy process to manage the selection of appropriate and inexpensive generators, maintenance schedule arrangements and moreover about plant planning due to electricity demand growth. Medium term load forecasting (MTLF) is needed to address the needs for optimizing the supply of electricity. A simple classification of electrical load forecasting has been presented in this paper [1] and from this paper load forecasts can be divided into 3 types based on the forecast period, namely short-term load forecasting, medium-term load forecasting, and long-term load forecasting. Short-term forecasting aims to optimize the supply of electricity in a system by paying attention to aspects of system reliability and also economic aspects. The duration generated in this forecast is generally in half an hour or 1 hour, this forecast lasts for 1 day to weekly. Meanwhile, medium-term load forecasting aims to plan fuel reserves and other primary energy and to confirm unit commitments. The results of this forecast are generally in the form of peak load and average daily consumption. Forecasting duration from 1 month to 1 year ahead. And lastly is the long- term load forecasting, this forecast is used for expansion of the construction of new power plants and also transmission system used for power evacuation. The duration for this forecast lasts for the estimated load from annual to 10 years. Researchers over the past few decades have used many methods to perform load forecasting. Load forecasting methods are generally divided into two, namely statistical methods and artificial intelligent methods. Statistical forecasting methods are used if there is data that can be taken from research in the form of time series data[2]. Time series data is past data consisting of the results of measurements or observations arranged in series according to the order of time. The more data that can be obtained for several types of statistical methods, the more accurate the forecasting results will be. Statistical methods consist of several methods including time series techniques, linear regression [3], linear autoregressive methods [4], exponential methods [5] and also includes the stochastic time series method[6]. Also included are Autoregressive Integrated Moving Average (ARIMA) [7, 8], Autoregressive (AR) [9]. The effect of sudden changes in environmental or sociological variables such as changes in