Assessing the Performance of Top-Ranking Indian Companies Using Self-Organising Map and Discriminant Analysis

Economic information extracted on financial ratios is extensively used by researchers for many purposes. To facilitate the grading of top-ranking companies, the financial information of public and private sector companies rated as the best with reference to net sales, published by Business Standard, was considered for the period from 2011 to 2016. Out of numerous ratios, 19 financial ratios that had different notions of the objectives and significant meaning in the literature were sieved carefully. Factor analysis is initiated first to uncover the structural patterns underlying financial ratios. The factor scores were then used in k-means cluster analysis to prune the original data. The self-organising principle is used first on a map of 10 by 5 neurons with the hexagonal topology for the neighbourhoods. The weight-vectors of these 50 neurons are then subjected to the k-means algorithm that yielded only three meaningful clusters for each year. By considering the centroids of the groups obtained from the weight-vectors, the iterated discriminant analysis is carried out on the original data until 100% classification is achieved. This analysis is carried out for the data relating to the period from 1994 through 1999. It is observed that there are only three groups persistent in each year. By assuming that the rates of increase in the group means are constants from one year to the next, the group means are increased by the corresponding rates to mine the data for the years from 2012 onwards to inherent the patterns from the previous year. It is also interesting to note that the clusters obtained could be arranged according to the magnitude of their group means on the ratios, thus permitting the groups to be identified on the basis of their performance. Finally, the groups were identified as companies belonging to Grade A, Grade B and Grade C in that order, which exhibit the behaviour of High performance, Moderate performance and Low performance.