FROM IDENTIFICATION TO BUDGET ALLOCATION: A NOVEL IT RISK MANAGEMENT MODEL FOR ITERATIVE AGILE PROJECTS

Transcription

1 FROM IDENTIFICATION TO BUDGET ALATION: A NOVEL IT RISK MANAGEMENT MODEL FOR ITERATIVE AGILE PROJECTS 1 AHDIEH KHATAVAKHOTAN 1 NAVID HASHEMITABA 1 SIEW HOCK OW khotan@iwa.um.edu.my nhtaba@iwa.um.edu.my how@um.edu.my 1 Department of Software Engineering, Faculty of Computer Science and Information Technology, Univerity of Malaya, 50603, Kuala Lumpur, Malayia ABSTRACT The increaing advancement in information technology indutry i compatible with the changing nature of the world. While oftware i important for all facet of the modern world, oftware development itelf i not a perfect proce. Agile oftware development method have recently emerged a a new and better way of developing oftware a compared to traditional methodologie. Thee method are developed to make prompt change in a contantly changing world.the more modern the technology and the method are, the newer the rik that mut be properly controlled and managed. Therefore, focuing on rik management proce in Agile proect will lead to applicable reult. Due to the high failure rate in IT proect, increaing the focu on rik management eem eential. Thi reearch i carried out to explore and determine a new rik management model for iterative Agile proect. The model tart by rik identification of each iteration baed on the hitorical data and end with rik budget etimation and allocation. Therefore, a comprehenive IT rik management proce will be performed baed on Agile approach. The rik etimation i by mathematical formulae which make the calculation defenible and reliable. KEY WORDS Agile approach, Rik management, Budget etimation, Rik identification 1. INTRODUCTION The management of rik conidered a a key dicipline for proect management [1]. According to the latet reearch done in 2007, 32% of oftware proect fail [2]. Failing a proect mean not implementing the planned program. The rik management role i vital to decreae the failure rate. Although, different rik management method are argued in the pat year, but till there i not a formal reliable model for rik management [3]. Epecially in Agile proect rik management mut be done promptly and accurately to avoid rework. The main point of the Agile manifeto i reponding to change [4].A the oftware development proce become fater by Agile approach, rik management proce mut be adapted with that. Therefore all phae of rik management hould be done completely but promptly in and Agile proce [5]. Agile approach i more human baed rather than proce baed [6]. Therefore, the maor rik i people in rik management proce.

2 In thi paper the main iue i to preent a model which provide a wide cope from rik identification to rik budget etimation and allocation. The aim i to explore the exiting rik of the Agile method and preent an appropriate model for a comprehenive rapid rik management plan. 2. THE STEPS OF THE MODEL Step 1- Identifying the rik by categorizing them. Step 2- Analyzing each rik in any iteration to determine: I. Which tak are affected by the mentioned rik II. Computing the weight of rik by: How relevant each tak i (low to high) If removal of the rik i neceary for ytem accuracy baed on hitorical data III. Calculating the Rik Factor amount (F1) IV. Prioritizing the known rik baed on the Biaed Rik factor (F2) Step 3- Obtaining the cot of each rik Step 4- Calculating the rik amount (F3) Step 5- Calibrating of rik according to the ued Agile method Step 6- Computing the requeted rik budget (F4) Step 7- Calculating the real budget for each rik (F5) Step 8- Monitoring the rik removal proce Step 9- Analyzing the acquired reult in order to apply in the next iteration 3. AN EXPLANATION OF THE MODEL In tep 1, rik identification will be completed and a preliminary lit of potential rik baed on previou iteration (except for the firt iteration) will be provided. Thi i inpired by the hitorical data meaning [7]. Step 2 ha four tage. Stage I, log the affected tak by the aforementioned rik in the preliminary lit, while the econd tage (II) conider two important feature for any of the ri relevancy and removal neceity. Thee feature facilitate the rik identification to prioritize the ignificant rik and update the initial lit. By entering the tage III the original formula (F1) help to calculate the rik factor amount. Step 2 finihe by prioritizing rik by biaed rik factor (F2) RISK FACTOR AND TOTAL RISK AMOUNT CALCULATION Each Rik Factor ha a definition and a value. h how the amount of rework hour (overtime) for rik i, to be performed by emi_killed peron in iteration (k i the counter variable for ). i the amount of rework hour (overtime) for all rik, to be done by emi_killed people in iteration For Rik i, Iteration :(F1) H

3 RikFactor i, k 1 k 1 k 1 k 1 k *[ h k *[ H h H ) ) ) ) h H h) h) h) h) ] ] Where N i a Normalized Coefficient, it i calculated according to the ranking of involved people in the needed hour (high_killed, killed, and emi_killed). The Normalized Coefficient, which i mentioned above, i ued in the method ut for firt iteration and have to be calibrated baed on the hitorical and the environmental data that formed the following metric for the other iteration. The meaning of metric and Calibrated Normalized Coefficient which i hown by n ymbol are preented in next ection. 1 emi _ killed ( N 2 killed ( ) 3 high _ killed ( h) For each identified rik i, in iteration, the criteria of RikFactor calculation will be a follow:for the firt iteration, the um will be ued ut baed on the experience had been acquired from imilar proect. Therefore, rik enacting in thi model will be from the econd iteration. To calculate rik factor in iteration (n), all rik in firt iteration to the iteration (n-1) will be conidered. Jut the only difference, i that like COCOMO etimation model, which wa preented by Boehm, the lat iteration will have more weight than the firt one. Hence, rik management normalized hour in each iteration, will be multiplied by the number of that iteration and in thi way will gain weight in accordance with the cloene of that iteration to the current iteration. The variable i ued for the diplay of iteration (the weight of each iteration i K) Metric and it Uage for the Calculation of n The metric ued in thipaper i baed on the number of line of code () produced by each category of people. Hence, the ratio of the number of line written by high_killed, killed and emi_killed to the number of line will be normalized quotient (calibrated N). Thi amount i obtained on average, for three categorie of high_killed, killed and emi_kiled people, a the following formulae: Iteration : h h n h Iteration : n Iteration : n Where i the Number Of People involved in proce (h,,, alo h, mean the number of high_killed people in iteration.

4 (h) N (h) () N() ( N( Calculation of weight Function and BiaedRikFactor Rik calculation in a hitorical tream i the amount of Rik Factor in every running or development of a ytem that i multiplied by relevancy and neceity factor. In a hitory baed model of Cheng and Jiang [35] the only factor which i conidered i relevancy; however, in the uggeted model here, the weight of any factor w, i a function of two variable n, r (neceity and relevancy both). To make it impler, thi function i defined a dicrete function between 0 and r 1, n 1 w(r,n)= 1... r 1, n r 0, n 01 Rik : i Iteration : BiaedRikFactor w ( r, n ) i i * RikFactor i (F2) BiaedRikFactorI (F3) 3.2. RISK REQUESTED BUDGET AND REAL COST ESTIMATION After accomplihing the rik identification and analyi tage, the rik budget requet and the allocation policy would be initiated. RequetedR ikbudget =RikCot RikCot Expected _ Cot * TotalRikA mount Expected _ Cot H * Fee H * Fee H * Fee Fee i the payment for each hour of human work. For more accuracy in thi formula, conidering the cot baed on ranking of people i important. For rik management, firt the um of needed budget and cot of each rik i calculated. The budget can be computed by the following formula: (F4) RequetedRikBudget= RikCot t te The rik management ha to be applied baed on the aigned budget, not the requeted budget, becaue, thee two budget are not often equal. The real allocated (given) budget to any rik i could be calculated by the following formula: (F5) h h

5 BiaedRikFactor i BudgetRik i * AignedRi kbudget TotalRikAmount 4. CASE STUDY In order to verify the model, a cae tudy ha been completed. Thi cae tudy ha depicted accurate rik cot and etimation of an IT company which produce web-baed and window-baed oftware for oil and ga indutry [8]. The relevant information by one of the Agile method (Scrum) in two year i gathered from thi company and applied to thi formula-baed model. Therefore, comprehenive rik management plan including p budget etimation and allocation i produced for the aforementioned company. 5. SUMMARY Agile method are almot newfound in IT era. Therefore, there are unknown rik in thi emerging method. The Agile development proce i alo fater than conventional model, which make the rik management more difficult to be applied. In an iterative Agile approach the proce rik occur and repeat and intenify gradually. The impact will be amplified promptly. The reaon i the rapid iteration in Agile approach. Thi paper preented a novel model for rik management. Neverthele, thi theoretical model would not be ufficient and rik management proce i not applicable unle the rik cot and etimation complete. Therefore, correpondent rik budget etimation formulae are added to make the model complete. By completing each iteration, the hitorical data become more accurate and real. Finally, to verify the model a cae tudy of an IT company with Agile infratructure examined the correctne and accuracy of the formulae. REFERENCES [1] Kutch, E., & Hall, M. (2009). The rational choice of not applying proect rik management in information technology proect.proect Management Journal, 40(3), doi: /pm [2] Vinon,J. 2007,How many of the IT proect fail? Too many.knowledge Jolt, p1. [3] Yuwei, S. (2007). Today' Solution and Tomorrow' Problem: The BuineProceOutourcing Rik Management Puzzle. California Management Review, 49(3), [4] Bec Kent,et al. Manifeto for The agile Software Development. February [5] Nerur.S,Mahapatra.R, and Mangalara.G, 2005, Challenge of Migrating to The Agile Methodologie; Communication of the ACM (48:5), P [6] Pichler.S and Pichler.R, 2005,The Agile Rik / The Agile RewardSoftware Development.San Francico: Vol.13, Iue 4, P [7] Jiang, H,Chang,C., Jinchun.X., Cheng, S. 2007, A Hitory-Baed Automatic Scheduling Model for Peronnel Rik Management, Compac, pp , 31t Ann. Int. Computer Software and App. Conference - Vol. 2. [8] Richard Mann,C. Augut 2005, An Exploratory Longitudinal Cae Study of Agile Method in a Software Company, Department of Computer Science, Univerity of Calgary, Alberta, Canada, ISBN