Analyzing the features of startup failure in financial sector via spectral clustering

Financial sector development is essential to the economic growth of our country. Various features that contributes to the Startup failure in finance affects the growth of the economy either directly or indirectly. The study of those features plays a crucial role in reducing Startup failures and increasing the sustainability of startups for entrepreneurs, investors, and policymakers. This paper focuses on the identification of features that are attributed to the startup failure data and groups the data based on similar characteristics. In this article, higher dimensional data relevant to the startup failure in the financial sector are collected. Using principal component analysis higher dimensional data is reduced to the lower dimensional and the corresponding eigen spectrum indicating the most influential feature is also presented graphically. metrics like the silhouette score is used to estimate the optimal number of similar groups following pre-processing and transforming the similarity matrix. Using the results obtained from the group of clusters with analogous features, the percentage of failures corresponding to the dominant startup features are estimated and the results are graphically illustrated. This analysis provides fruitful information to the future entrepreneurs to build their business.