Projects risk analysis based on data mining

با توجه به ماهیت پویایی و موقتی محیط های پروژه ای که در بردارنده سازمانها و استخدامهای موقتی اکثر پرسنل کلیدی می باشد، جریان قوی تولید دانش سازمانی مشهود است ولیکین بدلیل همین ماهیت موقتی در صورت نبودن سازوکار متفکرانه برای مدیریت دانش پروژه بسیاری از اطلاعات و دانش ایجاد شده به فراموشی سپرده می شوند. این در حالیست که مشخصه های نامبرده سازمانهای پروژه ای با حداکثر ریسکهای زمانی، هزینه و تکنولوژیک در بخشهای قراردادی، مهندسی، تامین تجهیزات، اجرا و راه اندازی رو به رو هستند که بایستی در سازمانهای اصلی برای تامین سودآوری، کیفیت و به هنگام بودن پروژه ها مد نظر و توجه قرار گیرند.

در این مقاله کشف و استخراج دانش از طریق داده کاوی به عنوان یکی از فرایندهای آماده سازی داده ها جهت استفاده در فرایند مدیریت ریسک پروژه های اجرایی مورد استفاده قرار گرفته است و در بر دارنده یک پژوهش علمی در زمینه کاربرد داده کاوی برای استفاده در فرایند تصمیم گیری و تصمیم سازی مدیریت پروژه در راستای شناخت ، خوشه بندی، دسته بندی، آنالیز و رتبه بندی ریسکهای پروژه از طریق کشف قوانین و الگوهای پنهان در داده های تاریخی موجود در پایگاه های داده سازمانهای پروژه محور برای استخراج دانش پنهان ویا ناشناخته در این زمینه است.

در این مقاله ابتدا با یک مجموعه مرور ادبیات به دو مقوله توجه شده است؛

· نقش آنالیز ریسک در مدیریت پروژه و مراحل اجرای آن

· تعادریف دانش داده کاوی و حیطه و دامنه کابرد آن

در ادامه به تعریف ساختار شکست ریسک، چیستی آن، چگونگی ایجاد آن، و کابرد آن در مدیریت ریسک در پروژه پرداخته شده است، ایجاد این ساختار شکست ریسک خود یکی از منابع دانش سازمانی محسوب می شود که یکی از بهترین و کاراترین روشهای مدیریت این دانش، ایجاد دانش بانک پروژه ها می باشد که با استفاده از ابزارهای داده کاوی مخصوصا متدهای دسته بندی، خوشه بندی و درخت تصمیم و ترکیبی از این روشها دست به ایجاد ساختار شکست ریسک پروژه ها به عنوان مبنا و چارچوبی برای اجرای مدیریت دانش پروژه ای می باشد. در این سیستم داده ها و اطلاعاتی که به صورت روزانه در پروژه های مختلف ایجاد می شود با برچسب گذاری ریسک در یکی از دسته های پیشین یا یک خوشه جدید دسته بندی شده و بنا به روشهایی که در پیش آمد نسبت به برآورد نسبت خطر و احتمال رخداد آن اقدام می گردد.

با توجه به گسترش ساختارهای اطلاعاتی و ارتباطاتی تحت وب این امکان فراهم می گردد که اطلاعات به هنگام و بر خط به روز آوری شده و کلیه دست اندر کاران پروژه با دیدی واقعی در مورد استراتژی های اجرایی پروژه بطور هوشمندانه ای تصمیم گیری نموده و گام بردارند

Abstracts:

In this article, the discovery and extraction of knowledge through data mining as one of the processes to prepare data for use in Projects risk management process is used And contain a scientific research in the field of data mining application for use in project management decision-making and decision supporting process with recognition, classification, analysis and ranking of project risks through discovery hidden rules and pattes in the historical Project-oriented organizations data base for Hidden or unknown Knowledge extraction in this area.

Keyword: Risk management, Risk analyzing, Data Mining, Risk Breakdown Structure (RBS), Risk Breakdown Matrix

Introduction:

Data mining including using sophisticated data analysis tools to discover existing pattes and unknown relationships between the data in broad volume. These tools include statistical models, mathematical algorithms and leaing methods. Consequently, data mining in addition to data collection and data management is containing some analysis and predictions.

One of The most important applications of data mining is project risk analysis based on organization historical available data. For Implementation a risk management model based on data mining, Data mining implementing process should be conforming to the risk management steps.

In This paper some of the data mining applications and methods for risk analysis as one of the most important data mining tasks has been reviewed

Literature Review

Knowledge of data mining is one of the top ten sciences in last century. Data mining or in the other perspective knowledge discovery from data bases is extraction of non-obvious information on potentially useful data that have been previously unknown.

The subject includes some of technical methods such as clustering, data summarization, classification rule, networks relationships, change analysis and irregularity discovering

this knowledge Is using different sciences such as statistics, mathematics and artificial intelligence to fulfill the needs of a new response, including prediction and discovery of hidden pattes in data and is recognized as a bridge between statistics science, artificial intelligence, patte recognition, machine leaing and visual representation.

Risk management is introduced by project management institute, as one of the twelve main levels of « Project Management Body of Knowledge »

According to definition of this Institute, project risk management phases are divided to risk identification, risk measurement, providing response (reaction against risk) and risk control.

In this definition, project risk management is the term «all processes related to identifying, analyzing and responding to any uncertainty, including maximize the results of desired events and minimize the results of adverse events is».

Project risk management

Project is a unique activity with restrictions in resources includes human resources, capital, raw materials and time to achieve certain goals.

There are Variable factors in future processes that are unknown for managing project, they called risk.

Thus according to the market development of inteational project, the risk management theory has been developed.[1]

Although the risk management process will be used during the project implementation phase.

But the Risks of the project specified in all stages of the project life cycle.

As theory, any risk that is determined during the project life cycle is to be managed as part of a risk management process.

Without proper risk management processes, to achieve desired goals in cases of financial, time and quality are faced with danger.

This process includes activities that are set to reduce the probability and severity of impact of each risk.

The risk management process is used to ensure all the risks are identified, documented and the probability of occurrence has been increased or decreased.

The definition of Risks is such: any event which is likely negative effect for the project’s ability to achieve the preset goals.

The project management considers to the following five key processes during project implementation:

· Project Risk Identification

· Project Risk Recoding and Ranking

· Project Risk Decreasing Action Nomination

· Project Risk Decreasing Action Assignment for implementation and Supervision

Closing the project risks

The risk management process is done to ensure that all risks come formally identified, Rated, supervised and the occurrence of them has been prevented or their effect has been reduced.

Risk management process contains the phases of identification, assessment, control and feedback.

Duty of each phase is affected by one of the following factors: technology, process and human recourse as the core of any organization and risk resources, each phase of the risk management process must separately performs for each of these critical boundaries.[2]

The relation between data mining and risk management

Data mining technology enables the organizations and companies to exploit the data and information assets and use them for supporting the process.

The usages of this knowledge are considered by scientists in different fields and the results of it encourage them to wider use of data researching.

Project Management and Risk Management are the new fields of using the Data mining by scientists.

Risk management process is the method that the project risks (in the scope, project output and resources ...), are identified, ranked and managed by it.

The risk has several concepts.

Referring to the definitions, each researcher defined it according to his reasons and discussions.

In the most definitions, two aspects of risk namely uncertainty and loss are clearly mentioned, But the third aspect, namely Choosing, is usually mentioned implying, choosing means how to attend it, these three conditions are the fundamental foundations of risk and the base of deeper survey of it.

Two problems should be solved For Implementation of risk management theory:

1- Dividing the whole project from start to finish achieving the risk components in project.

2- Determining the transition matrix of risk components between each of the risk factors.

For the first problem can use the experiences, environment analysis or method WBS in project management and for the second problem can use the on data analyzing based on concepts and records. [3]

Risk Breakdown Structure is a source-oriented hierarchical structure for classification of the risks in the project.

Risk Breakdown Structure is one of the most important tools in identifying risk and actually is can be said that identifying the risk without preparation of risk Breakdown Structure is a very hard and irrational act, therefore it’s suggested to prepare the risk breakdown structure in the first step of risk identification.[11]

Creating the risk breakdown structure (RBS), a hierarchical structure of project risks analysis has been studied. As that creating a dynamic contrast between the RBS components in the lowest levels of risk and WBS in the lowest level of Work Packages through a matrix of bilateral based on the intensity and importance of each component on each product can be the "Risk Breakdown Matrix" (RBM) way. [11]

The role of Data mining in risk analysis of project

Data mining is a process that can automatically discovers the hide information, categories and the information related to statutory approach.

Using database technology, data base is available in abundance contain a lot of hide valuable information.

Therefore, how to discover the information is the most of the genuine achievements of data researching.

In other words, Data mining is a process discovers unknown and valuable knowledge based on abundant historical data, using the idea of data researching. Transition matrix for risk factors can be made as it is shown in Figure 1;

1-Determine the risk information by WBS components or other methods and input them in database for future use.

2 - Information refinement by filtering and remove the noises.

3 - Analysis of data for Data mining to get the transition matrix of risk elements, for these purposes the following two steps is needed:

3-1 - Categorize of risk scenarios in database.

3-2 - Determine the possibility of transition of each mode and categories. [3]

The output of risk analysis is risk matrix, the location of risk elements is determined based on incidence probabilities and the level of effect intensity and through it the key risk indicators (KRI) will be achieved.

These indicators are the quantitative measurement that is the tools for hazard alert and verification of risk incidence.[2]

Hence the use of developed methods in Data mining is better than previous statistical methods because empirical research in the risk management field has shown that the output accuracy of past methods such as Partial variance methodology would be reduced by the non-real information. [4]

Project risk management process helps the financial project supporters and the project teams to make conscious decisions about alteative solutions. Risk management encourages the project team to choose proper methods for minimizing the negative effects on scope, cost and plan of the project and also on the crisis management. [11]

To identify the various performance and financial risk factors, levels and signs of waing, should use the categorizing methods that are used as the main methods in Data mining methods such as logical regression, task analysis, cluster analysis, SOM, C&RT, CHAID and etc. Decision tree algorithm that is one of the section methods, due to understand and visual implement easiness can be used [5] the decision tree give the permission of imagine and visualize of a series of decision to functional signs easily. [6]

Different methods of Data mining for risk analysis

Decision tree method can obtains through the Data mining to discover information of researches and the probability of incident of decision elements can calculated by distribution of risk factors. Firstly we should supply the decision tree for risk analysis through the historical data and Data mining is one of the new theories for determine the hidden relations among data sets, Data mining is used for making the decision trees, classifying, clustering and etc. [1]

Anticipant phased methods predict the most probable event in the nearest time and they are manager assistants in decision-making and risk management, also discovering hidden rules with high confidence in them can be the points of secrets in the Information for managers.

Mainly the experts do not know how the risk indicators should be combined to get the final risk ranking. Therefore, the individual risk indicators make the inputs of neural network that calculates the final ranking of risk, this network is trained to specify the available samples of database through the known risks, if the laws would be right and the samples of database do not have a lot of disorder, the ranking of calculated risk is almost equal to the actual risk. [8]
Events risk analysis tree is the method of decision tree and has been a logical conclusion is that analyze all results created with historical events to get the system quantitative and qualitative characters. [1]

However improved algorithm is not often sufficient, restriction of Data mining can remove only through the integration of expert knowledge and encoding through the available methods, Mostly Data mining algorithms create some structures or models which are able to collate with human decision makers due to their complexity. [8]

Methodology

Considering the temporary and dynamics nature of the project environments that contain most temporary organizations and key persoel employment, Strong flow of knowledge production is evident, but However due to the temporary nature If lack of intelligent mechanisms for projects knowledge management, many information and knowledge created to be forgotten. While the characteristics of those project organizations, with maximum risk in time, cost and technologies in contract, engineering, Procurement, construction and commissioning sectors are faced that in the main organization for providing Profitability, quality and on time being Projects takes into consideration

One of the best and most efficient methods for managing this knowledge, is projects Knowledge Bank that using data mining tools, particularly classification, decision trees and clustering and a combination of these methods to create projects risk break down structure as the basis and framework for implementing the project knowledge management.

In this system data and information that each day in various projects will create by risk labeling is classifying in former classes or in new clusters and the ways that went the amount and probability of its occurrence is Determine.

Considering expanding Web-based information and communication structures this is possible provided that information are on line updates and all project members with real vision about the executive project strategies to take intelligent decisions and take actions.

Conclusion

according to the mentioned items, hierarchical structure of the risk elements would be create through the mutual matrix and risk breakdown that gives the key sources risk elements in high levels for risk managers and for MS analysis in lower levels in the form of Data mining risk management. This structure is a hierarchical structure that describes the resources of project risks. These tools allow different levels of project risk to be increasing partial. At high levels all risks are "project risks". The structure in the first level can be broken to the main sources of risks, such as technical risks, commercial risks, management risks and exteal risks. Each of the main range can be broken to the lower levels (for example the technical risks can be broken to the technology, performance, reliability, and interference and etc. risks) in the lowest level, single risks identified through the related source of resources. In the other hand the input data updates through the input data to organization database and updated analysis of these data for making the matrix through the updated Data mining and we always see the last changes and decides based on current data.
انجمن مهندسی و مدیریت ساخت ایران (ICEMA) Iran Construction Engineering Management Association
Suggestions

Considering the requirements for the reduction of risk costs in project, the use of mode methods of Data mining and knowledge discovery systems through to the member and abundance is required. Considering the provisional and dynamic essence of organizations based on projects that unfortunately cause loss or forgetfulness of the vast majority of data and hidden knowledge of project categories in mind of project team, compile and create a knowledge bank based on knowledge management concepts and techniques of Data mining such as clustering, especially in the Downloads, refinement, recognition, classification and data analysis will be a lot useful.

1. Resources

2. Cunbin Li, Jianjun Wang. The Application of Tree Risk-Element Model Based on Rough Set Data Mining in Project Risk Management. The Sixth Wuhan Inteational Conference on E-Business － Engineering Technology Track. 2007

3. R. S. Kenett and O. Raphaeli. Multivariate Methods in Enterprise System Implementation, Risk Management and Change Management. IJRAM (2005)

4. LI Cun-bin and WANG Jian-jun ,Model of generic project risk element transmission theory based on data mining, J.Cent. South Univ. Technol.,(Springer) 2008

5. J. Aerić and E. Jurun and S. Pivac. MULTIVARIATE RISK-RETURN DECISION MAKING WITHIN DYNAMIC ESTIMATION. Economic Analysis Working Papers.- 7th Volume – Number 11 (2007)

6. S. Koyuncugil, and N. Ozgulbas, A Data Mining Model for Detecting Financial and Operational Risk Indicators of SMEs, PWASET(PROCEEDINGS OF WORLD ACADEMY OF SCIENCE, ENGINEERING AND TECHNOLOGY) VOLUME 36, DECEMBER 2008

7. Stefano Pisani and Paola De Sisti, Risk Analysis applied to Tax Evasion using data mining methodology, 2007 K. Jugdev and G. Mathur and T. S.

8. Fung. Project management assets and their relationship with the project management capability of the firm. Inteational Joual of Project Management 25 (2007)

9. Heie Daniels and Han van Dissel, RISK MANAGEMENT BASED ON EXPERT RULES AND DATA-MINING: A CASE STUDY IN INSURANCE, ECIS, 2002

10. Sh. Hsieh, J. H. Chou, Forecasting of Value at Risk (VAR) by Cluster Method in Chinese Stock Market. Joual of Money, Investment and Banking. 2008

11. G. Fox, D. Ebbeler and E. Jorgensen. THE USE OF CLUSTER ANALYSIS TECHNIQUES IN SPACEFLIGHT PROJECT COST RISK ESTIMATION. American Institute of Aeronautics and Astronautics

12. D. Hillson, S. Grimaldi and C.Rafele. MANAGING PROJECT RISKS USING A CROSS RISK BREAKDOWN MATRIX. Risk Management (2006) 8, 61 – 76