Performances Measures of Top-Ranking Companies Using Self Organizing Maps
G. Manimannan1*, G. Saravanan2, G. Senthil Kumaran3
Citation :G. Manimannan, G. Saravanan, G. Senthil Kumaran, Performances Measures of Top-Ranking Companies Using Self Organizing Maps International Journal of Scientific and Innovative Mathematical Research 2019 , 7(5) : 4-9.
The research paper is to study whether the classical classification techniques, specifically the use of Self Adaptive neural network models based on Kohonen�s theory, make easier to evaluate the performance of the chosen industrial sectors which are ranked according to their net sales from 2010 to 2015 with unknown group information. Out of numerous ratios that could be constructed using various financial parameters, eleven financial ratios were chosen that had different notions of the objective and significant meaning in the literature. Initially, factor analysis is applied to hidden structural patterns underlying financial ratios. The factor scores were extracted, and then used for k-means clustering techniques to prune the original database. The reduce data is then subjected to the final analysis using the proposed algorithm which results in the task of suitable grades to the companies. The method chosen also offers a way of visualization position of the company�s performances through topological two-dimensional self organizing maps.