Financial Feasibility Study of New Electricity Installations with Line Expansion Based on Machine Learning

The financial feasibility study is one of the important processes of serving new electricity installations with line expansion at company as it must consider infrastructure construction investment to be connected according to prospective customer requests. The financial feasibility consists of Net Present Value, Internal Rate of Return and Payback Period, which lead to a recommendation of the feasibility of a project. Some research that discusses feasibility studies, finance, and machine learning have been done but they are separate and are not interconnected with each other. Some are still limited to the calculating methods based on manual formulas, especially for a complex situation where we must consider big datasets. To have a reliable, fast, and measurable financial feasibility study, a system that can perform the calculation process to be able to determine decisions quickly and based on supervised data learning is required. However, there are several supervised machine learning models, that can be chosen to implement such feasibility studies. This paper presents a comparison of several supervised learning models, including elastic net, ridge, lasso regression, random forest, and linear regression. The aim of the comparison is to provide better and complete information for the management of Indonesia's state-owned electricity company (PLN), when they want to implement the investment assessment of new electric installations as a response to potential customer requests. Using this investment assessment decision makers have good information and recommendations for appropriate, quick, and measurable decisions.