Paper Title
Rainfall Prediction: Accuracy Enhancement for Myanmar Using Machine Learning

The designation of this project is to enhance accurate rainfall data to support weather and disaster prediction by using advanced technological tools and to save people by reducing forecasting time and cost. Since heavy rainfall can affect many disasters, rainfall is an important weather variable for meteorology, hydrology, and climatology. For the economy and lifetime of humans, heavy precipitation prediction, a cause for natural disasters like floods and drought, could be a major responsibility for the meteorology and hydrology department. Accurate prediction is useful for people to take preventive measures. There are two types of prediction: short-term rainfall prediction and long-term rainfall prediction, mostly shortterm predictions can give us accurate results, and building a model for long-term rainfall prediction is a challenge. The accuracy of rainfall is extremely important for agricultural countries like Myanmar. The prediction of precipitation using machine learning techniques may use regression and multi-class classification in this research. The fifteen years of monthly historical weather data collected from two stations were used to train as input for learning and testing the models for regression and four weather stations in Yangon were used for classifications. Observed weather attributes such as Pressure, Temperature, Humidity, Wind Direction, Wind Speed, Rainfall, and Dew Point Temperature are used as input and the output is Rainfall. According to the correlation between weather parameters, different combinations of input weather parameters were used to build the models. In this case study, the purposes are not only for accuracy but also for time. After assessing the accuracy and processing time results, Decision Forest performed better for regression and Neural Network is better in multi-class classification although the process running time is not too much different by 30 minutes. Keywords - Rainfall Prediction, Accuracy, Machine Learning, Microsoft Azure, Decision Forest, Neural Network