Hennie Daniels

Methodological and practical aspects of data mining

Abstract We describe the different stages in the data mining process and discuss some pitfalls and guidelines to circumvent them. Despite the predominant attention on analysis, data selection and pre-processing are the most time-consuming activities, and have a substantial influence on ultimate success. Successful data mining projects require the involvement of expertise in data mining, company data, and the subject area concerned. Despite the attractive suggestion of ‘fully automatic’ data analysis, knowledge of the processes behind the data remains indispensable in avoiding the many pitfalls of data mining.

Application of MLP Networks to Bond Rating and House Pricing

Feedforward neural networks are receiving growing attention as a data modelling tool in economic classification problems. It is well known that controlling the design of a neural network can be cumbersome. Inaccuracies may lead to numerous problems in the application, such as higher errors due to local optima, overfitting and ill-conditioning of the network, especially when the number of observations is small. In this paper we provide a method to overcome these difficulties by regulating the flexibility of the network, and by rendering measures for validating the final network. In particular, a method is proposed to equilibrate the number of hidden neurons based on 5-fold cross-validation. In the validation process, the performance of the neural network is compared with a linear model using 5- fold cross-validation. In both case studies, the degree of monotonicity of the output of the neural network, with respect to each input variable, is calculated by numerical differentiation. The outcomes of this analysis are compared to what is expected from economic theory. Furthermore, a special class of monotonic neural networks and a corresponding training algorithm are developed. It is shown in the second case study that networks in this class have less tendency to overfitting than ordinary neural networks. The methods are illustrated in two case studies: predicting the price of housing in the Dutch city of Den Bosch; and the classification of bond ratings.

Estimating structural exchange rate models by artificial neural networks

No theory of structural exchange rate determination has yet been found that performs well in prediction experiments. Only very seldom has the simple random walk model been significantly outperformed. Referring to three, sometimes highly nonlinear, monetary and nonmonetary structural exchange rate models, a feedforward artificial neural network specification is investigated to determine whether it improves the prediction performance of structural and random walk exchange rate models. A new test for univariate nonlinear cointegration is also derived. Important nonlinearities are not detected for monthly data of US dollar rates in Deutsche marks, Dutch guilders, British pounds and Japanese yens.

Monotone and Partially Monotone Neural Networks

In many classification and prediction problems it is known that the response variable depends on certain explanatory variables. Monotone neural networks can be used as powerful tools to build monotone models with better accuracy and lower variance compared to ordinary nonmonotone models. Monotonicity is usually obtained by putting constraints on the parameters of the network. In this paper, we will clarify some of the theoretical results on monotone neural networks with positive weights, issues that are sometimes misunderstood in the neural network literature. Furthermore, we will generalize some of the results obtained by Sill for the so-called min-max networks to the case of partially monotone problems. The method is illustrated in practical case studies.

Explanation of Exceptional Values in Multi-dimensional Business Databases

In this paper, we describe an extension of the OnLine Analytical Processing (OLAP) framework with causal explanation, offering the possibility to automatically generate explanations for exceptional cell values. This functionality can be built into conventional OLAP databases using a generic explanation formalism, which supports the work of managers in diagnostic processes. The central goal is the identification of specific knowledge structures and reasoning methods required to construct computerized explanations from multi-dimensional data and business models. The methodology was tested on a case study involving the comparison of financial figures of a firm’s business units. The findings suggest improved decision-making by managers because the current tedious and error-prone manual analysis process is enhanced by automated problem identification and explanation generation. It is also noted that this novel methodology has general utility for decision-support systems, for example, for automated diagnosis in the financial and accountancy domain.

Qualitative dynamics and causality in a Keynesian model

Abstract In this paper we present a formalism to describe economic dynamics in a qualitative way. This formalism is a modification of an existing algorithm for qualitative simulation as proposed by Kuipers. It is demonstrated that the framework of qualitative dynamics can clarify economic reasoning without using any quantative data. Especially causal arguments that sometimes mysteriously occur when economists implicitly mix static and dynamic models, can be understood in a formal way. Furthermore, we bring together the lines of thought recently established in the field of artificial intelligence and the results of qualitative economics that can be found in earlier papers. A simple Keynesian model serves as an example throughout this text.

Decision trees for monotone price models

Abstract.In economic decision problems such as credit loan approval or risk analysis, models are required to be monotone with respect to the decision variables involved. Also in hedonic price models it is natural to impose monotonicity constraints on the price rule or function. If a model is obtained by a “unbiased” search through the data, it mostly does not have this property even if the underlying database is monotone. In this paper, we present methods to enforce monotonicity of decision trees for price prediction. Measures for the degree of monotonicity of data are defined and an algorithm is constructed to make non-monotone data sets monotone. It is shown that monotone data truncated with noise can be restored almost to the original data by applying this algorithm. Furthermore, we demonstrate in a case study on house prices that monotone decision trees derived from cleaned data have significantly smaller prediction errors than trees generated using raw data.

A general model for automated business diagnosis

Abstract In this paper we develop a formalism to support diagnostic reasoning in the domain of business and finance. A theoretic description of the process of diagnosis of company performance, and the implementation thereof is outlined. A new concept of explanation comprises the basis of our framework and enables us to deal with both qualitative and quantitative information in diagnosis. The system was tested on a case-study involving the comparison of nine firms operating in the mechanical engineering industry. Comparison of textbook analysis and model output shows that our system is able to produce the correct diagnostic analyses.

2010 Special Issue: Comparison of universal approximators incorporating partial monotonicity by structure

Neural networks applied in control loops and safety-critical domains have to meet more requirements than just the overall best function approximation. On the one hand, a small approximation error is required; on the other hand, the smoothness and the monotonicity of selected input-output relations have to be guaranteed. Otherwise, the stability of most of the control laws is lost. In this article we compare two neural network-based approaches incorporating partial monotonicity by structure, namely the Monotonic Multi-Layer Perceptron (MONMLP) network and the Monotonic MIN-MAX (MONMM) network. We show the universal approximation capabilities of both types of network for partially monotone functions. On a number of datasets, we investigate the advantages and disadvantages of these approaches related to approximation performance, training of the model and convergence.

Causal reasoning and explanation in dynamic economic systems

Abstract In this paper, we develop techniques for qualitative reasoning in economic systems. It is shown that qualitative economic reasoning can be formalised to the extent that qualitative behaviour and associated explanations can be obtained which correspond to economic reasoning put forward by economists. Existing qualitative reasoning techniques are capable of generating allpossible behaviours of an economic system out of which many have no satisfactory economic explanation. In order to constrain this intractable branching of behaviour, a heuristic filter based on the casual dependencies in the economic system selects those behaviours which are meaningful from an economic point of view.