This is the help for STAR. XOMO [1] is the initial tool which inspired this one. These two tools both aim at minimizing Effort, Defects, and Threat based on the COCOMO II [2], Coqualmo, and Threat Models [3] for given project parameters that are based on these models. However, the two tools take two differet approaches for tackling the same problem. While XOMO uses Tar3 treatment learning [4], STAR uses simulated annealing [5], combined with best or rest discretization (bore) and a linear back select algorithm to show the best treatments that can be chosen (i.e. the ones discarded last in the back select process). The temperature of the SA algorithm is set to exp(A*k/kmax), where kmax is the maximum number if iterations allowed, k is the number of iterations so far, and A is a constant that is chosen. The following are options allowed with this version of STAR: Option Description ------ ----------- -sim or -SIM Used in the form of "-sim [value]" to determine simulation count during the backselect process -ex or EX To enable extreme sampling. -run or RUN To indicate run number. -al or -AL To set the weight for the COCOMO II effort model (default = 1). This is followed by the desired weight in the form of -al [value]. -be or -BE To set the weight for the COQUALMO defects model (default = 1). This is followed by the desired weight in the form of -be [value]. -ga or -GA To set the weight for the threat model (default = 1). This is followed by the desired weight in the form of -ga [value]. -de or -DE To set the weight for the months model (default = 1). This is followed by the desired weight in the form of -de [value]. -rd or -RD To set the weight between reliability and defects model (default = 1.8). This is followed by the desired weight in the form of -rd [value]. -cf or -CF To set the cooling factor used in Simulated Annealing (default = -10). This is followed by the desired value in the form of -cf [value]. Note that this needs to be a negative number. -n or -N Bore for worst: This allows the user to get the worst treatments instead of the best. -h or -H To display this help file and exit. -l or -L To enable logging and post processing: Logging as in keeping a log of the results of the simulated annealer, and post processing being the back select and simulations. -er or -ER To enable energy ranking. The default ranking method is Baysian. -lc or -LC To output individual simulations within back-select runs -or or -OR To output individual back-select runs -f or -F Used in the form of -f [project name], this indicates the project being analyzed. This option must be included for STAR to work. Also, both the *.values and *.ranges for the project files must be in the STAR_projects folder. -b or -B Used in the form of "-b [value]". Sets the ratio that bore will use for discretization. -e or -E Used in the form of "-e [value]". Sets the min energy threshhold for simulated annealing. -k or -K Used in the form of "-k [value]". Sets the max number of iterations for simulated annealing. -pol or -POL Used to specify the back-select policy in the form of -pol [policy name]. -st or -ST Used to specify the Hoh in strategy in the form of -st [strategy name]. -bf or -BF Better/Faster -fc or -FC Faster/Cheaper -bc or -BC Better/Cheaper There are several files included with STAR. The following is a brief description of each: File Name Description --------- ----------- graph This is a bash script that uses generated simulation data to produce plots in the command line shell. graphPNG This is a bash script that uses generated simulation data to produce plots in *.png format. /eg This folder contains sample bash scripts of running STAR. /STAR_projects This folder contains project files. Note that for a single project, the ranges and values files are in the form "[Project Name].ranges" and "[Project Name].values". /policies Contains all the policy files. Inorder to get started simply make. All the needed files will be in $HOME/STAR/, while the executable will be in $HOME/bin. The following is an example: STAR -pol strategic -l -f ground After running STAR, plots will be dumped onto the command line. Note: gnuplot is needed for this. References: ----------- [1] T. Menzies. RST milestone 1.1.5.9: Applying trade space analysis to recommend CEV/CLV options for SW capabilities and development of processes and toolshttp://menzies.us/pdf/06xomo202.pdf [2] B. Boehm, E. Horowitz, R. Madachy, D. Reifer, B. K. Clark, B. Steece, A. W. Brown, S. Chulani, and C. Abts. Software Cost Estimation with Cocomo II. Prentice Hall, 2000 [3] R. Madachy. Heuristic risk assessment using cost factors. IEEE Software, 14(3):51–59, May 1997 [4] T. Menzies and Y. Hu. Data mining for very busy people. In IEEE Computer, November 2003. Available from http://menzies.us/pdf/03tar2.pdf [5] S. Kirkpatrick and C. D. Gelatt and M. P. Vecchi, Optimization by Simulated Annealing, Science, Vol 220, Number 4598, pages 671-680, 1983.