Application of generalized principal component analysis to classification of price levels of markets in Ghana

Abstract

The study determines the main dimensions along which to classify specific markets as extremely high-priced, high-priced, medium priced, low priced and those that are extremely low-priced. For this purpose, secondary data is obtained on prices of 19 food items in 91 Markets in non-consecutive years from 2008 to 2015 in all regions of the country from the Statistical, Research and Information Directorate of the Ministry of Food and Agriculture. This data constitutes a multivariate dataset with nineteen variables and 455 observations. Generalized Principal Component Analysis, which is a procedure for data classification and summarization, is considered appropriate for analyzing this high dimensional dataset. An appropriate index is designed to identify the various levels of the prices and eigen-analysis among others further informed the choice of the technique. It is found that there are indeed extreme priced markets in Ghana. The results reveal that the classification of the markets is not influenced by their location. Another important observation is that root and tubers and cereals are major components of local food items. In order to reduce the disparity in price levels there is the need for concentration to be given to the production of these two types of local food items for economic wellbeing of Ghanaians