Εφαρμογή στατιστικών μεθόδων πολυμεταβλητής ανάλυσης σε φορολογικά δεδομένα επιχειρήσεων
Application of multivariate statistical methods in enterprise taxation data
Λόκκα, Ευδοκία Θ.
In the present MSc. thesis, methods of Multivariate Analysis are applied on Greek industrial business and financial data. These financial data are used by tax authorities to assess companies’ financial position and viability. These estimates may become useful for both companies and the state itself since they provide clues for a possible need of preventive measures that will favour the avoidance of adverse future economic situations like (e.g. bankruptcy). Key financial ratios of efficiency, capital structure, activity and liquidity are used for predicting business’ default. The predictive power of these indicators is examined by the use of methods of Multivariate Analysis such as Logistic Regression, Probit Analysis, Principal Component Analysis and Clustering methods. All methods, apart from Clustering, exhibit a good fit to the data. Logistic Regression revealed that the index “Equity to Assets” owns the best power for predicting a business’ default.