Extreme Value Analysis
Weather forecasting is a broad subject that has been in application since time immemorial. There are very many components of weather that contribute to its forecasting. One of the valuable components of weather is rainfall. This paper makes the use of rainfall trend to make weather forecasts. The prime objective of the paper is to use the analysis of the trend of rainfall as portrayed by the data to predict the extreme weathers changes within the next two decades. The paper begins with an introduction that outlines the main objective of the analysis and the methods used to achieve this objective. It outlines the application of Extreme Value Theory (EVT) to come up with a conclusive analysis that will help predict weather changes based on previous rainfall trends on the data. The next section is on data mining. The data used in this analysis was sourced from the National Meteorological Department in the country Kenya’s capital. The data was mined from their open public website. The next section outlines the methodologies used in the analysis. The method used in this case is the Block Maxima Method through the application of Generalized Extreme Value (GEV) distributions. The mathematical manipulations and computation using this method would require extensive statistical software due to the nature of the data. The paper then proceeds to give the data analysis based on the Block Maxima methodology. The results of this analysis would help discuss the future extreme and rather rare rainfall trends that would be duly conclusive based on the accuracy and precision of the data and its analysis.