Clustering Firm Financial Performance Using Neural Networks: Experimental Results in Urban Areas

This paper aims at using neural networks – and especially self‐organizing maps (SOMs) – to analyse and cluster firm financial performance. Our study addresses a large set of financial data of companies settled in the urban area of Genoa, a city in Northern Italy. We use SOMs to both cluster the selected firms depending on their performance and to visualize emerging patterns about firm features. SOMs are well‐known as a visual clustering method supporting complex data analysis and decision support. In our work, the proposed approach is applied to a database of companies settled in a specific urban area, with the aim to produce an effective visualization of firm clusters depending on their different performance profiles. These profiles are the further basis of performance analysison clusters of homogenous firms. SOMs permit to cluster firms depending onseveral features concurrently and therefore to create revealing clusters about the different firm performance profiles. Experimental results demonstrate the functionality of SOMs to explore large amount of data and to visualize and cluster the features of different groups of companies settled in the selected urban area. These results are useful to support decisions about local policies for territorial economic development.