Accuracy Vancouver Rain Forecast Based on KNN Algorithms

Abstract

Vancouver is a waterfront city with warm-summer Mediterranean Climate, which means the rainfall frequency is relatively high, so rain forecast can provide convenience to the society and our daily lives. Since uncertainty exists in weather forecasts, this study provides the K-nearest neighbors algorithm, which is a machine learning algorithm, as a possible simple version substitution for the numerical weather prediction model (NWP), and aims to answer the research question: what is the accuracy of the rain forecast in Vancouver based on the KNN algorithms? The research will use laboratory experiment and data analysis to answer the question. The algorithms will be coded and trained using the historical temperature data from the Vancouver International Airport Weather Station. After training, the current weather data will be input to the algorithms to get the result, and the actual weather will be recorded. Finally, the accuracy of the prediction will be calculated using the number of factually consistent forecasts over the total forecast number. The expectation of the answer is approximately 70%.

Biographies

My name is Leo Gao, and I am a first-year student in Vantage Science at UBC. I am interested in artificial intelligence and game design.

My name is Salman Alqallaf, and I am a first-year student studying science at UBC. I am interested in biology and medicine. My interest is how to combine machine learning with medicine.