Innovative Applications of O.R.
Using FSBT technique with Rough Set Theory for personal investment portfolio analysis

https://doi.org/10.1016/j.ejor.2009.03.031Get rights and content

Abstract

This study proposes a novel Forward Search and Backward Trace (FSBT) technique based on Rough Set Theory to improve data analysis and extend the scope of observations made from sample data to solve personal investment portfolio problems. Rough Set Theory mathematically classifies data into class sets. The class set with the most objects may generate one decision rule. The rules generated from RST are rough and fragmented, that are very difficult to interpret the information. An empirical case is used to generate more than 85 rules by the RST method in comparison with FSBT method which only generated 14 rules. This result can show our proposed method is better than traditional RST method based on class sets that contain the most objects. Much of human knowledge is described in natural language. It is a very important thing to convert information from computer databases into normal human language. Sample data taken from features with the same backgrounds are used to compile different portfolios that investment companies and investment advisors can employ to satisfy the investor’ needs. The method not only can provide decision-making rules, but also can offer alternative strategies for better data analysis. We believe that the FSBT technique can be fully applied in research on investment marketing.

Introduction

A well-designed financial plan can help to achieve good asset allocation and meet customer needs. Indeed, asset management is closely tied to personal experience and behavior. For example, in an inflationary market, investment capital decreases as personal expenses increase.

In recent years, research concerning attitudes towards personal wealth has increased. Moreover, there has been an increase in the public’s interest in wealth creation, which is perhaps best reflected in the burgeoning literature on financial matters. With regard to financial hardship, research suggests that individuals’ past experiences usually affect their attitudes towards making investments. However, there is now a trend toward relying on financial services as an effective tool for increasing personal wealth.

Assessing financial services is a complex task because the consumer has to evaluate the features of product before it has actually been consumed. For example, when consumers adopt a particular investment plan, they cannot really evaluate the quality of the outcome until the investments mature. Hence, they may be forced to make a decision before assessing the result.

Among financial institutions, there is a growing preoccupation with customer retention and relationship marketing – essentially understanding the behavior of consumers after the initial purchase has been made, and focusing on how to foster a profitable relationship with the clients. This is crucial to the success of financial institutions, and the approach recognizes the ongoing nature of the relationship and the longevity of many financial products.

Prior to making purchase decisions, consumers can gather information from a variety of sources. The factors that affect decisions associated with personal asset allocation are the risk level and return revenue of investment products, the timing strategy (the time of buying and selling products), and the portfolio. Additionally, individuals may have different investment needs, reflecting their personal backgrounds and life experiences as well as their individual personalities. These factors can make personal asset allocation decisions very difficult.

To date, there have been relatively few attempts to develop models that explain consumer decision processes specifically in the context of financial services. A considerable amount of theoretical and empirical work exists relating to how businesses make financial decisions for clients generally, rather than how they personalize their services for a single customer. From the buyer’s perspective, the decision-making process can be roughly divided into four components: problem recognition, information search, evaluation of alternatives, and purchase decision. The decision-making process may also be affected by one’s past life experience and personality. The process of selecting an appropriate investment portfolio can be divided into two stages. The first stage starts with observation and experience and ends with beliefs about the future performance of available securities. The second stage starts with relevant beliefs about the future performance of various investment products and ends with the choice of portfolio (Markowitz, 1952). Many papers have been published on this topic; specific subjects include the behavior of financial services consumers (Harrison, 2003), management of personal finances (Teichman et al., 2005), retirement plans (Hanisch, 1994), personalization of intelligent financial decision support systems (Palma-dos-Reis and Zahedi, 1999), development of an intelligent system for personal and family financial services (Chieffe and Rakes, 1999), and assessment of the impact of customer satisfaction and relationship quality on customer retention (Hennig-Thurau and Klee, 1997). A number of studies focus on quantification of the problem by streamlining all the parameters and applying statistical tools to analyze the data. Rough Set Theory (RST) is a method for discovering knowledge from ambiguous information. Developed by Pawlak, 1982, Pawlak, 1984, RST is a rule-based decision-making technique that can handle crisp data sets and fuzzy data sets, without the need for a pre-assumption membership function, as required by fuzzy theories (Zadeh, 1965). Membership is not the main concept in the Rough Set Theory. Rough sets can deal with indiscernibility knowledge and fuzzy sets deal with vagueness knowledge. The results of the Rough Set Theory are composed of classification and decision rules which derived from a set of examples. More comparison details between fuzzy set and rough set theories can refer the paper of Walczak and Massart (1999). RST has become a hot topic because of its application to knowledge discovery in real-world databases or warehouses.

However, RST generates many rough rules that are not easy to apply. In this empirical study, there are more than 85 rules generated by the RST method. Thus, we propose a new search method called Forward Search and Backward Trace (FSBT), which extends the search process after the RST technique finds the decision class set with the most objects and only generate 14 rules in FSBT method. That set represents the desired answer or model, and we use its components as criteria in a forward search to find matching objects (called Target objects) that have the same components as the decision class set with the most objects. We can then discover a consumer’s consumption behavior from those components (the customer’s backgrounds). It is a very important thing to convert information from computer databases into normal human language and much of human knowledge is described in natural language. Complete information for a rule is an important thing for Decision Maker to do the decisions.

The rules generated from RST are rough and fragmented, that are very difficult to interpret the information. However, the rough set approach can be considered as a formal framework for discovering facts from imperfect data. The advantage of FSBT is that, compared to classical RST, it can cull more information from the same backgrounds/personalities data sample. Classical RST only generates rough rules that have the best lower approximation rate to perform an analysis. This result for data utilization is very inefficient. However, the FSBT method can expand the observation scope and reduce data utilization inefficiency. Financial advisors can derive more investment combinations with FSBT applications. In reality, investment experts can review a customer with the personal characters/backgrounds, estimate which investment portfolio is suitable for the customer and give adequate suggestions by the FSBT method. Moreover, the investment experts can perform special analysis and provide further service to the special customers (the outlier objects) supplied by the FSBT method.

In this paper, we focus on identifying different types of information derived from the same characters/backgrounds to create more personalized investment portfolios for satisfying the investors’ needs and helping them increase their personal wealth. We use RST for feature classification, and then try to determine the priority of investments at different levels. The results may help investment companies or investment advisors propose suitable investment products to fulfill their customers’ needs. There have been relatively few studies on the use of RST for personal investment analysis. For this study, we designed a questionnaire to investigate personal investment portfolios. Using real cases of investors in Taiwan as the basis of the empirical study, we tried to determine the participants’ priorities when they make personal investment portfolio decisions. The questionnaire considered the factors that affect decision-making, such as sex, age, and the number of family members; monthly income (Harrison, 2003, Plath and Stevenson, 2005); the nature of the investment products; and participants’ basic data, which may serve as the basis for understanding their needs.

The results of the study demonstrate the efficacy of the proposed FSBT method, and identify three types of personal investment portfolio: conservative portfolio, moderate portfolio, and aggressive portfolio. Investors who choose conservative portfolios are generally single college graduates under the age of 29, with less than four years work experience and a monthly salary below US$900. The majority of people with an aggressive portfolio are married, female college graduates under the age of 39, with 15–19 years work experience and a monthly salary below US$2424. Those with a moderate portfolio are generally single, under 29 years of age, with a monthly salary between US$900 and US$2424. The investment priorities for conservative portfolio consumers are bank deposits, insurance policies, and houses. The priorities of those with moderate portfolios are bank deposits, houses, and insurance, while those with aggressive portfolios prefer houses, bank deposits, and stocks.

The remainder of this paper is organized as follows. In Section 2, we discuss personal investment portfolios. In Section 3, the methodology of FSBT combined with RST is described. In Section 4, we consider a real case of a personal investment portfolio to explain the process of the proposed method. Then, in Section 5, we present some concluding remarks. To distinguish between RST and FSBT, classical RST is denoted as RST throughout the paper.

Section snippets

Personal investment portfolio

In an investment portfolio, the factor with the most impact on investment revenue is asset allocation. Hence, investors should modify their investment portfolios depending on their needs at regular intervals. An investment portfolio is composed of many investment products that must be managed effectively in order to increase personal wealth.

Investment products can be divided into two categories: risk products and non-risk products. Risk investments include stocks, mutual funds, foreign

RST and FSBT methods

We now introduce RST and FSBT methods and explain the procedures used to analyze personal investment portfolios. Section 3.1 describes the history of RST, Section 3.2 details the RST algorithms, and Section 3.3 discusses the FSBT method. Decision rules can be approached by enhancing the approximations and accuracy of the decision rules of the classification and checking the quality of the classification, all of which can be used to extract suitable rules and describe the information system. In

Empirical study of the FSBT method compared with the classical RST method for personal investment portfolio selection

Under highly competitive conditions, the best way to access markets and enlarge market share is to acquire needed information from potential customers through well-designed surveys. A successful business not only fulfills customers’ needs, but also designs business strategies and/or measures to improve the firm’s performance.

The questionnaire for personal investment portfolio analysis is discussed in the following.

Conclusions

We have used the FSBT method to study the asset management market in Taiwan. Based on the survey results, we identified the following types of personal investment portfolio holder:

Conservative: (1) less than 29 years old, (2) college graduate, (3) single, (4) under 4 years work experience, and (5) monthly salary under NT$30,000 (US$900).

Aggressive: (1) less than 39 years old, (2) female, (3) married, (4) a college graduate, (5) 15–19 years work experience, and (6) monthly salary under NT$80,000

Acknowledgement

Thanks very much to the reviewers, and Editor, B., Slowinski, for extensive, useful comments which helped to improve the presentation of this article. For the sake of the paper limits to be published, a supplementary material can refer from the web site: http://cid-c0798d7de4c5b827.skydrive.live.com/self.aspx/FSBT%20paper.

References (32)

  • Z. Pawlak

    Rough classification

    International Journal of Man–Machine Studies

    (1984)
  • Z. Pawlak

    Rough sets, decision algorithms and Bayes’ theorem

    European Journal of Operational Research

    (2002)
  • D. Plath et al.

    Financial services consumption behavior across Hispanic American consumers

    Journal of Business Research

    (2005)
  • M. Quafafou

    α-RST: A generalization of rough set theory

    Information Sciences

    (2000)
  • P. Sawicki et al.

    Technical diagnostic of a fleet of vehicles using rough set theory

    European Journal of Operational Research

    (2009)
  • J. Shyng et al.

    Rough set theory in analyzing the attributes of combination values for the insurance market

    Expert System with Application

    (2007)
  • Cited by (38)

    • Incremental updating of rough approximations in interval-valued information systems under attribute generalization

      2016, Information Sciences
      Citation Excerpt :

      The uncertain concept and knowledge hidden in an information system can be described by the lower and upper approximations in RST. Nowadays, many rough set based data analysis approaches have been successfully applied in medical diagnosis [12], financial analysis [4,38], industrial control [32], investment decisions [1,29,30] and many other fields. Data analyzed by RST are commonly interpreted and presented through the information system, which includes objects, features (attributes) and attribute values [26].

    • Combined soft computing model for value stock selection based on fundamental analysis

      2015, Applied Soft Computing Journal
      Citation Excerpt :

      The classical RSA was proposed by Pawlak [28], in an attempt to tackle the indiscernibility and ambiguity of complex data sets. Although the classical RSA achieved major success in discovering useful knowledge in various applications—such as personal investment portfolio analysis [29], prediction of financial distress and credit assessment decision—the RSA was constrained in dealing with the non-ordered data of the classification problems [30]. However, in practice, many problems require handling of data by using preference-ordered attributes.

    • A rough set-based corporate memory for the case of ecotourism

      2015, Tourism Management
      Citation Excerpt :

      It is suitable for processing qualitative information that is difficult to analyze using standard statistical techniques (Heckerman et al., 1997). RST has been applied in numerous fields, for example, supplier selection (Bai & Sarkis; 2010), decision support systems (Degang & Suyun, 2010), investment portfolios (Shyng, Shieh, Tzeng, & Hsieh, 2010), marketing (Cheng, Chen, & Lin, 2010), EEG analysis (Ningler, Stockmanns, Schneider, Kochs, & Kochs, 2009), multiple criteria classification (Dembczyński, Greco, & Słowiński, 2009; Zhang, Shi, & Gao, 2009), expert systems (Shao, Chu, Qiu, Gao, & Yan, 2009), databases (Wang & Wang, 2009), time series analysis (Sarkar, 2006; Teoh, Cheng, Chu, & Chen, 2008; Yao & Herbert, 2009), image compression & segmentation (Mushrif & Ray, 2008; Petrosino & Ferone, 2008) and boolean reasoning (Pawlak & Skowron, 2007). In RST, a general decision table for the representation of the relationship between condition attributes and decision attributes is used.

    • A hybrid stock trading system using genetic network programming and mean conditional value-at-risk

      2015, European Journal of Operational Research
      Citation Excerpt :

      They propose an interactive multiobjective optimization method based on so-called Dominance-based Rough Set Approach (DRSA). Shyng, Shieh, Tzeng, and Hsieh (2010) propose a novel Forward Search and Backward Trace (FSBT) technique based on Rough Set Theory to improve data analysis and extend the scope of observations made from sample data to solve personal investment portfolio problems. Tay and Shen (2002) demonstrate that rough sets model is applicable to a wide range of practical problems pertaining to economic and financial prediction.

    • Do impression management tactics and/or supervisor-subordinate guanxi matter?

      2013, Knowledge-Based Systems
      Citation Excerpt :

      This method is based on the indiscernibility relation by a dominance relation in the rough approximation of decision classes. The basic concepts of DRSA are described as follows [4,9,12,14,22–24,28,38,45,49,50,52]. In DRSA, data are frequently presented in the form of a decision table.

    View all citing articles on Scopus
    View full text