Predictions of Future Aspects of the Rainy Season Using Simple and Multiple Linear Regression Analysis- A Case Study of Chingóme Mission Daily Rainfall Data in Zambia
Urban Nchimunya Haankuku
This article demonstrates the point and interval predictions of the dependent variables Y using both simple and multiple linear regression analyses for the given independent X variables. The primary methodology was analysis of quantitative data collected at Chin’gombe mission, northern part of Zambia, weather station for a period of 25 years. The article begins by justifying why a particular approach was used for analysis by testing the available data for randomness. Since, time trends where not evident, a classical approach was adopted which involved the construction of models that reflect the available data as closely as possible. A distribution with two parameters was preferable for greater flexibility, hence, the truncated exponential distribution with two unknown parameters was investigated instead of other distributions such as; lognormal, gamma, or weibull. But the predictions obtained were not particularly informative for agricultural planning, water management, and designing purposes. The article also shows methods of analysis of daily rainfall data using both simple and multiple linear regression analysis and demonstrates how the models derived can be of direct use in agricultural planning, water management and designing. The correlations between the onset and end of rainy season dates were also investigated. The results obtained showed that there is no correlation between the onset and end of rainy season.
Full Text: PDF