EXPLORATORY DATA ANALYSIS AND MULTIVARIATE STRATEGIES FOR REVEALING MULTIVARIATE STRUCTURES IN CLIMATE DATA

摘要

This paper is on data analysis strategy in a complex, multidimensional, and dynamic domain. The focus is on the use of data mining techniques to explore the importance of multivariate structures; using climate variables which influences climate change. Techniques involved in data mining exercise vary according to the data structures. The multivariate analysis strategy considered here involved choosing an appropriate tool to analyze a process. Factor analysis is introduced into data mining technique in order to reveal the influencing impacts of factors involved as well as solving for multicolinearity effect among the variables. The temporal nature and multidimensionality of the target variables is revealedin the model using multidimensional regression estimates. The strategy of integrating the method of several statistical techniques, using climate variables in Nigeria was employed. Rof 0.518 was obtained from the ordinary least square regression analysis carried out and the test was not significant at 5% level of significance. However, factor analysis regression strategy gave a good fit with Rof 0.811 and the test was significant at 5% level of significance. Based on this study, model building should go beyond the usual confirmatory data analysis (CDA), rather it should be complemented with exploratory data analysis (EDA) in order to achieve a desired result.