Abstract:
"ri Lanka is currently facing an acute challenge in meeting its rapidly growing electricity demand
amid an economic crisis and constrained capacity for energy import and export. The conventional
methods for forecasting energy supply and demand, as well as for identifying and evaluating
potential energy sources, fall short in addressing the complex, multi-dimensional energy landscape
of Sri Lanka. This project aims to develop a predictive model that offers a more accurate, efficient,
and sustainable solution for energy management in Sri Lanka, thereby addressing the country's
urgent need for reliable and cost-effective energy solutions.To tackle this problem, a Python-based
predictive model was developed using advanced computational techniques, including machine
learning algorithms. The project involved comprehensive data pre-processing to ensure highquality inputs, followed by the selection and tuning of appropriate machine learning models such
as regression analysis and ensemble methods. The model was designed to integrate various
economic, environmental, and consumption variables to forecast future electricity demand and
identify the most efficient and cost-effective energy sources for generation. This approach also
incorporated renewable energy potentials, aiming to provide a holistic view of energy management
for the country.
The developed model was evaluated using historical data and various data science metrics, such as
Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-squared values, to assess
its accuracy and reliability. The results demonstrated a high level of predictive accuracy, with the
model successfully forecasting future energy demands and highlighting the optimal mix of
renewable and conventional energy sources. These findings offer actionable insights for
policymakers and contribute significantly to the discourse on sustainable energy management in
Sri Lanka"
