Abstract
This article is a technical note that describes a conventional agricultural regionalization method and its problems, using the Hokuriku region as an example. The method is a multivariate analysis that combines factor analysis and cluster analysis. Finding agricultural regions requires the following 12 steps; 1) selecting variables, 2) preliminary calculations, 3) omitting variables, 4) weighting factor scores, 5) exponentiating factor scores, 6) selecting a scale of statistical unit, 7) interpretation of factors, 8) map design, 9) displaying maps for factor analysis, 10) interpretation of clusters, 11) displaying a map for cluster analysis, and 12) regionalization. We then discussed these problems. (i) How many variables should be selected? (ii) Which statistical unit should be selected? (iii) How can the calculation be adjusted? (iv) How can regions be classified subjectively?
