Non-numerical Bankruptcy Forecasting Based on Three Trends Values - Increasing, Constant, Decreasing

Abstract

There is a broad spectrum of different BM (Bankruptcy Models). However, complex bankruptcies are unique, vaguely known, interdisciplinary and multidimensional. These are the key reasons why sufficiently large sets of examples are not available It is therefore often prohibitively difficult to make forecasts using numerical quantifiers and traditional statistical methods. BMs development suffer from IS (Information Shortage). IS eliminates straightforward application of traditional statistical methods based on information rich environment; that is on the law of large numbers. Artificial Intelligence has developed different tools to minimise IS related problems. Trend reasoning is one of them. It is based on the least information intensive quantifiers There are four different trends i.e. qualitative values and their derivatives: plus/increasing; zero/constant; negative/decreasing; any value / any trend. The paper studies BMs represented by models based on EHE (Equationless Heuristics). An bankruptcy example of EHE is - If Selling of Assets is increasing then Satisfaction of Creditors is increasing. Such verbal knowledge items cannot be incorporated into a traditional numerical model. No quantitative quantifiers, e.g. numbers, are used in this paper. The solution of a trend model M(X) is a set S of scenarios where X is the set of n variables quantified by the trends. All possible transitions among the scenarios S are generated. An oriented transitional graph G has as nodes the set of scenarios S and as arcs the transitions T. An oriented G path describes any possible future and past time behaviour of the bankruptcy system under study. The G graph represents the complete list of forecasts based on trends. An eight -dimensional model serves as a case study. Difficult to measure variables are used, e.g. Level of Greed, Political Influence. There are 65 scenarios S and 706 transitions T among them. A priory knowledge of trend reasoning is not required.There is a broad spectrum of different BM (Bankruptcy Models). However, complex bankruptcies are unique, vaguely known, interdisciplinary and multidimensional. These are the key reasons why sufficiently large sets of examples are not available It is therefore often prohibitively difficult to make forecasts using numerical quantifiers and traditional statistical methods. BMs development suffer from IS (Information Shortage). IS eliminates straightforward application of traditional statistical methods based on information rich environment; that is on the law of large numbers. Artificial Intelligence has developed different tools to minimise IS related problems. Trend reasoning is one of them. It is based on the least information intensive quantifiers There are four different trends i.e. qualitative values and their derivatives: plus/increasing; zero/constant; negative/decreasing; any value / any trend. The paper studies BMs represented by models based on EHE (Equationless Heuristics). An bankruptcy example of EHE is - If Selling of Assets is increasing then Satisfaction of Creditors is increasing. Such verbal knowledge items cannot be incorporated into a traditional numerical model. No quantitative quantifiers, e.g. numbers, are used in this paper. The solution of a trend model M(X) is a set S of scenarios where X is the set of n variables quantified by the trends. All possible transitions among the scenarios S are generated. An oriented transitional graph G has as nodes the set of scenarios S and as arcs the transitions T. An oriented G path describes any possible future and past time behaviour of the bankruptcy system under study. The G graph represents the complete list of forecasts based on trends. An eight -dimensional model serves as a case study. Difficult to measure variables are used, e.g. Level of Greed, Political Influence. There are 65 scenarios S and 706 transitions T among them. A priory knowledge of trend reasoning is not required.