Visual differences of the three indices wehave built are insignificant. Moreover, mean-squared deviations of one indexfrom another make: 1,46 for φ19 and φ22, 1,75for φ19 and φ35, 0,30 for φ22 andφ23. In case the objects are ranked in accordance with indices set bythe functions we have built, the mean deviation in object numbers set byφ19 and φ22 is equal to 1,1; φ19 andφ23 makes 1,5; φ22 and φ23 is 0,5.. In the first case 93 objects have the same position; in the second case 71,in the third case 149. Only 64 of 237 objects keep their positions in all threerankings. In this case the sensitivity of the indicator to measurement errorsis less than in the preceding cases. It might be expected proceeding from thefact that statistical characteristics of the built indicators in this case arebetter.
4.2. Comparing the results: indicators andcluster analysis
This section is to compare the resultsobtained by clusterization of properties reviewed above (IDS, IA, EP) measuredby three indicators each (SPSL, PCISL, PCESL for IDS; SI, RGI, FI for IA; GRP,UL, FI for EP), the results are>
Since the indicator has values within the interval [0, 100], let usclassify the set of objects into М>
where М is the number of objects,, N is the number of clusters.
Let us > > objects, and the rest by ) objects.
Let us >
The necessity of such comparative analysis of different methods toclassify regions in accordance with the indicators is determined by the factthat it is possible to use the indicators as discriminant functions indifferent ways (for instance, in accordance with three methods offered in thisstudy). The choice of the best method is possible only after a thoroughsubstantive analysis. The results of this section are auxiliary and are used torender such analysis more easy.
Материалы этого сайта размещены для ознакомления, все права принадлежат их авторам.
Если Вы не согласны с тем, что Ваш материал размещён на этом сайте, пожалуйста, напишите нам, мы в течении 1-2 рабочих дней удалим его.