UA   RU  
DonNTU   Masters' portal

Abstract

At the time of writing this abstract, the master's thesis has not yet been completed. Estimated completion date: May-June 2022. The full text of the work, as well as materials on the topic, can be obtained from the author or his supervisor after the specified date.

Content

Introduction

The original forms of insurance originated in ancient times. The most ancient rules of insurance that have come to us are set out in one books of the Talmud. On the island of Rhodes in 916 BC an ordinance was adopted, which presents a system for distributing damage in the event of a general accident. The principles applied in this document have been preserved to this day [1].

With the growth of cities and the emergence of large settlements, the risk of death or damage to property from fires and other natural disasters increased. People began to unite for joint actions to prevent danger, eliminate consequences, including economic measures. So, in 1310, an Insurance Chamber, which was carrying out operations to protect the property interests of merchants and craft guilds, was established in Bruges (Germany).

The USSR Council of Ministers resolution of August 30, 1984 played a significant role in the development of the insurance business. On measures for the further State insurance development and improving the quality of work of insurance authorities. The situation has changed dramatically due to the legalization of entrepreneurship in Russia, when commercial, financial and economic risks have become a daily reality for ten thousand businessmen.

The revival of the insurance market in the country was in the early 1990s. By Presidential Decree of January 29, 1992 state and municipal insurance companies are transformed into closed and open joint-stock insurance companies (ASBs) and limited insurance liability partnerships (LLP). The Russian Federation Law On Insurance (dated November 27, 1992) entered into force on January 12, 1993. In 1996, a government decree On priority measures for insurance market development in the Russian Federation was issued.

An insurance company (a historically defined social form of the insurance fund functioning) is a separate structure that carries out insurance policies and their maintenance.

1. Relevance of the topic

The demand for insurance services is predetermined by the fact that economic entities (legal entities and individuals) constantly face the threat of some adverse or even catastrophic events that lead to significant financial losses (death, illness or dismissal from work of a family member whose work was the main source of income; property loss from fire; car accident, etc.). It is almost impossible to cover these losses from current income; it is also very difficult to accumulate funds for this through deposit accounts. Insurance is the most profitable compensation for such losses, since its amount may be more than insurance premiums.

Every insurance company tries to improve the concept of customer service: to reduce the processing time of customer questionnaires; to make the conclusion of insurance contracts comfortable for both the insurance agent and the client. The other side of this concept is the desire of the insurance company to protect itself from material losses in the form of numerous insurance payments. As a result, there is a need to study potential customers before concluding insurance contracts.

Due to the fact that many nuances have appeared in the insurance business and this area is constantly developing and a large number of changes can often be made, an ordinary employee of the company will not always be able to take into account all the nuances and subtleties, as well as all the individual features that come with each new client. This is where the system that is planned to be implemented comes to the rescue.

2. Purpose and objectives of the study

The aim of the work is to improve the quality of customer service, reduce the survey time and reduce (reduce) the material losses of the insurance company by developing a DSS that provides clustering of customers taking into account their individual characteristics and characteristics and the formation of mutually beneficial contracts.

The main objectives of the study:

  1. Define criteria for classifying indicators of insurance companies;
  2. Analyze clustering methods in relation to grouping of indicators;
  3. Develop test questions and perform their criterion by the selected parameters;
  4. Develop a PPR.

3. The current situation of decision support systems

Today, the software development level of the DSS class is characterized by a well-developed theoretical basis and a very narrow scope of application.

The very concept of a DSS implies the use of significant amounts of data; however, the list of characteristics of an ideal DSS does not contain the most important characteristic of the relationship with constantly replenishing data sources [2].

DSS is used to analyze a large volume of heterogeneous data, which means that the problem of sufficiency and timeliness of data provision is the most important, since absence or inaccuracy of data distorts the analysis results. Thus, the choice of a DSS development strategy requires the implementation of many functions that, although they do not fall into the DSS concept, are necessary to support decision-making. These functions include collecting, processing and transmitting of information.

4. Block diagram of the DSS

The decision support system is a set of software tools that includes a library of various decision support algorithms, a database of models, a database, auxiliary and control programs. The management program organizes the decision-making process taking into account the specifics of the problem.

Figure 1 shows a high-level block diagram of the expert system. As can be seen from Figure 1, the system provides the use of all the necessary blocks that the DSS should have. In order to simplify interaction between the user and the software tools, the user is required to form queries on the forms for the presentation of input and output information by the display and explanation units of the solution.

Структурная схема СППР

Figure 1 - Block diagram of the DSS

Animation: 9 frames, 5 repetition cycles, 87 kilobytes

The input data analysis unit focuses the client according to a grouped set of factors. Within each cluster, the boundary of the factor values is thus determined, conversion algorithms are applied using the weighting factors of each factor and the resulting values are further refined by fuzzy methods.

The logical output block receives data on the analysis carried out with an explanation of the basis on which decisions were made..

At the output of the analytical system unit, it is necessary to obtain the degree of risk that an individual client may pose to an insurance company. Depending on this result, the system should also take into account the answers to the test questions and, as a result, on the basis of the rules provide advice to the insurance agent which actions should be taken [2-3].

The main task of insurance companies is to minimize the payment of insurance cases. At the same time, it can be deduced from the customer’s individual data, on the basis of many centuries-old statistical research, that the risk the client may represent and whether to omit certain items in the future insurance contract. Taking into account a number of seemingly unrelated facts could save millions of insurance companies. Special analysts of insurance companies are responsible for identifying these patterns and the relationships between attribute values of clients.

So, minimizing the risks that a new client may pose to the company can help to minimize the costs of payments to insurance companies. At the same time, it is necessary to obtain personal data in a way convenient for the client. Often, the practice of insurance activity shows that if a person knows that an employee of an insurance company will personally check the questionnaire the client can give false information about his identity on various psychological aspects. Taking a questionnaire on a computer can increase the credibility index of the client’s personal data and allow conducting the client analyst at the same time and receiving results and advice for the specialist who engages in signing of insurance contracts.

5. Problem statement

The object of computerization is the process of concluding insurance contracts and analyzing the risks that the client presents to the insurance company. The legal basis of insurance is Federal Law No. 4015-1 of 27.11.1992 On Insurance, which also reveals the economic essence of insurance.

According to this Law, insurance is a relationship to protect the property interests of individuals and legal entities upon the occurrence of certain events (insured events) at the expense of monetary funds formed from insurance premiums paid by them (insurance premiums).

The subject of the insurance company's activity may be the following types of financial services:

Indicators can be grouped into 3 basic groups:

The group of factors FINANCE determines the information parameters characterizing the financial well-being of the client:

The HEALTH group determines the level of current physical health, as well as whether the client's current activities are safe and whether there are life-threatening hobbies:

The CONTRACT group defines the attributes of the insurance contract:

Each of the base groups has a weight, as well as each attribute within the group has a weighting factor of significance, depending on which it is possible to judge how much this or that factor affects the overall picture of the client's assessment as a whole. Intelligent processing of grouped data will allow you to assess the degree of risk of each client – RISK. With a low level of risk, it is possible to make decisions more flexibly, and it is possible to expand the insurance contract.

The parameters of the client's characteristics are represented as a vector RISK = {FINANCE, HEALTH, CONTRACT}.

Each element of the RISK vector is also a vector:

FINANCE = {salary, count_auto, count_credits, count_houses, is_married, count_children, Annual_Client_payment};

HEALTH = {is_aids +, count_operations, count_cigarets, count_alco, passport_age, fithness_age, risk_profession, risk_hobby, driving_experience};

CONTRACT = {SUM, YEARS, TYPE, insured_losses}.

Based on the rules of the knowledge base, the process of calculating the importance of the influence of each factor, the input data vector, and the vector is formed indx_Risk[].

Target function:

Целевая функция

Restrictions on variables:

  1. 0 ≤ AGE ≤ 60.
  2. If FINANCE.Annual_Client_payment ≥ (0.1 * FINANCE.Salary), then increase TOTAL_RISK.

For each group within a certain type of insurance, there is a set of weighting factors. Within each group, the total index is determined, in accordance with the rule base. Then it is multiplied by the group index and summed to get the final value:

Формула итогового значения

where subgroupW [j] – the value of the intra-group weighting factor j is determined from the rule base depending on the value of the j factor when the client is surveyed.

Conclusions

As a result of the analysis of the goals and methods of clustering and building decision support systems, it was found that:

References

  1. Развитие страхования в России – Страхование сегодня. История страхования [Электронный ресурс] / В. Г. Ларионов, М. Н.  Скрыпникова – Электрон. текст. – [Россия, 2000]. – Режим доступа: https://www.insur-info.ru/history/press/d2451762.
  2. Глухова, Н. В. Теория принятия решений: учебное пособие /Глухова Н. В. – Ульяновск: Ульяновский государственный педагогический университет имени И.Н. Ульянова, 2017. – 50 c. – Электронно-библиотечная система IPR BOOKS: [сайт]. – URL: https://www.iprbookshop.ru/86329.html. – Режим доступа: для авторизир. пользователей.
  3. Доррер, Г. А. Методы и системы принятия решений: учебное пособие / Г. А. Доррер. – Красноярск: Сибирский федеральный университет, 2016. – 210 c. – Электронно-библиотечная система IPR BOOKS: [сайт]. – URL: https://www.iprbookshop.ru/84240.html.
  4. Кластеризация: алгоритмы k-means и c-means [Электронный ресурс] – 2009 – Режим доступа: http://habrahabr.ru/post/67078/.
  5. Реализация алгоритма k-means на С# (с обобщенной метрикой) [Электронный ресурс] – 2012 – Режим доступа: http://habrahabr.ru/post/146556/.
  6. Наследов А. IBM SPSS Statistics 20 и AMOS: профессиональный статистический анализ данных. – [Россия, Санкт-Петербург, 2013]. – Глава 21. Кластерный анализ.
  7. Кластерный анализ [Электронный ресурс] – StatSoft: Электронный учебник по статистике – Режим доступа: http://www.statsoft.ru/home/textbook/modules/stcluan.html
  8. Иерархическая кластеризация [Электронный ресурс]: Режим доступа: https://ranalytics.github.io/data-mining/102-H-Clustering.html
  9. Кластерный анализ – Википедия [Электронный ресурс]: Режим доступа: http://ru.wikipedia.org/wiki/Кластерный_анализ