Rsrat provides the package to evalute the software reliability from the fault data collected in the testing phase. Rsrat can use two types of data; fault-detection time data and its grouped data. The fault-detection time data is a sequence of time intervals of fault detection times (CPU time, etc). Also its grouped data is a sequence of the number of detected faults for each time interval (per a working day, per a week, etc). The reliability evaluation is based on the software reliability growth model with NHPP (non-homogeneous Poisson process).
You can install Rsrat from GitHub with:
install.packages("devtools")
devtools::install_github("SwReliab/Rsrat")Alternatively, you can use Remote to install Rsrat from GitHub
install.packages("remotes")
remotes::install_github("SwReliab/Rsrat")This is an example of the estimation of software reliability growth models from a fault data (tohma).
### load library
library(Rsrat)
### load example data
data(dacs)
### tohma is a grouped data
tohma
#> [1] 5 5 5 5 6 8 2 7 4 2 31 4 24 49 14 12 8 9 4 7 6 9 4 4 2
#> [26] 4 3 9 2 5 4 1 4 3 6 13 19 15 7 15 21 8 6 20 10 3 3 8 5 1
#> [51] 2 2 2 7 2 0 2 3 2 7 3 0 1 0 1 0 0 1 1 0 0 1 1 0 0
#> [76] 0 1 2 0 1 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 1 0 0 0 1
#> [101] 0 0 1 0 0 1 0 0 1 0 1
### Esimate all models and select the best one in terms of AIC
(result <- fit.srm.nhpp(fault=tohma))
#> Model name: lxvmin
#> omega loclog scalelog
#> 481.7029 -3.4642 0.6637
#> Maximum LLF: -316.2599
#> AIC: 638.5198
#> Convergence: TRUE
### Draw the graph
mvfplot(fault=tohma, srms=result)
rateplot(fault=tohma, srms=result)
The second example illustrates the estimation for two specified models.
### All models in the package
srm.models
#> [1] "exp" "gamma" "pareto" "tnorm" "lnorm" "tlogis" "llogis" "txvmax"
#> [9] "lxvmax" "txvmin" "lxvmin"
### Estimate two models and no select
(result <- fit.srm.nhpp(fault=tohma, srm.names=c("exp", "gamma"), selection=NULL))
#> $exp
#> Model name: exp
#> omega rate
#> 497.2912 0.0308
#> Maximum LLF: -359.8777
#> AIC: 723.7555
#> Convergence: TRUE
#>
#>
#> $gamma
#> Model name: gamma
#> omega shape rate
#> 483.52301 1.88475 0.06447
#> Maximum LLF: -319.5695
#> AIC: 645.139
#> Convergence: TRUE
### Draw the graph
mvfplot(fault=tohma, srms=result)
### Draw the graph (dmvf)
dmvfplot(fault=tohma, srms=result)
rateplot(fault=tohma, srms=result)
The third example shows the case where the fault data are fault detection data.
### fault-detection time data
#### Time intervals for all faults
#### The last value is a negative value, that indicates the time interval in which there is no fault detection after the last fault detection.
sys1
#> [1] 3 30 113 81 115 9 2 91 112 15 138 50
#> [13] 77 24 108 88 670 120 26 114 325 55 242 68
#> [25] 422 180 10 1146 600 15 36 4 0 8 227 65
#> [37] 176 58 457 300 97 263 452 255 197 193 6 79
#> [49] 816 1351 148 21 233 134 357 193 236 31 369 748
#> [61] 0 232 330 365 1222 543 10 16 529 379 44 129
#> [73] 810 290 300 529 281 160 828 1011 445 296 1755 1064
#> [85] 1783 860 983 707 33 868 724 2323 2930 1461 843 12
#> [97] 261 1800 865 1435 30 143 108 0 3110 1247 943 700
#> [109] 875 245 729 1897 447 386 446 122 990 948 1082 22
#> [121] 75 482 5509 100 10 1071 371 790 6150 3321 1045 648
#> [133] 5485 1160 1864 4116 -2526
### Esimate
(result <- fit.srm.nhpp(time=sys1[sys1>=0], te=-sys1[sys1<0]))
#> Warning in emfit(srm, data, initialize = TRUE, maxiter = con$maxiter, reltol =
#> con$reltol, : Did not converge to MLE by max iteration.
#> Model name: gamma
#> omega shape rate
#> 1.545e+02 6.357e-01 1.618e-05
#> Maximum LLF: -967.1074
#> AIC: 1940.215
#> Convergence: TRUE
### Draw the graph
mvfplot(time=sys1[sys1>=0], te=-sys1[sys1<0], srms=result)
The fourth example illustrates the case where the mvfs for all the models are drawn.
### Esimate and return all the estimated results
(result <- fit.srm.nhpp(fault=sys1g, selection=NULL))
#> Warning in emfit(srm, data, initialize = TRUE, maxiter = con$maxiter, reltol =
#> con$reltol, : Did not converge to MLE by max iteration.
#> Warning in emfit(srm, data, initialize = TRUE, maxiter = con$maxiter, reltol =
#> con$reltol, : Did not converge to MLE by max iteration.
#> Warning in emfit(srm, data, initialize = TRUE, maxiter = con$maxiter, reltol =
#> con$reltol, : Did not converge to MLE by max iteration.
#> Warning in emfit(srm, data, initialize = TRUE, maxiter = con$maxiter, reltol =
#> con$reltol, : Did not converge to MLE by max iteration.
#> $exp
#> Model name: exp
#> omega rate
#> 4.295e+03 3.347e-04
#> Maximum LLF: -192.5611
#> AIC: 389.1221
#> Convergence: FALSE
#>
#>
#> $gamma
#> Model name: gamma
#> omega shape rate
#> 5.400e+02 1.799e+00 8.543e-03
#> Maximum LLF: -182.2326
#> AIC: 370.4651
#> Convergence: TRUE
#>
#>
#> $pareto
#> Model name: pareto
#> omega shape scale
#> 1179.22 56.39 44141.81
#> Maximum LLF: -193.7977
#> AIC: 393.5953
#> Convergence: FALSE
#>
#>
#> $tnorm
#> Model name: tnorm
#> omega mean sd
#> 150.45 60.97 26.72
#> Maximum LLF: -173.955
#> AIC: 353.91
#> Convergence: TRUE
#>
#>
#> $lnorm
#> Model name: lnorm
#> omega meanlog sdlog
#> 6188.497 8.270 1.839
#> Maximum LLF: -184.3571
#> AIC: 374.7142
#> Convergence: FALSE
#>
#>
#> $tlogis
#> Model name: tlogis
#> omega location scale
#> 153.37 62.34 16.17
#> Maximum LLF: -172.6565
#> AIC: 351.313
#> Convergence: TRUE
#>
#>
#> $llogis
#> Model name: llogis
#> omega locationlog scalelog
#> 392.6774 4.9056 0.5364
#> Maximum LLF: -181.6148
#> AIC: 369.2296
#> Convergence: TRUE
#>
#>
#> $txvmax
#> Model name: txvmax
#> omega loc scale
#> 208.49 62.51 38.96
#> Maximum LLF: -177.5718
#> AIC: 361.1435
#> Convergence: TRUE
#>
#>
#> $lxvmax
#> Model name: lxvmax
#> omega loclog scalelog
#> 18401.836 11.147 4.134
#> Maximum LLF: -186.8055
#> AIC: 379.611
#> Convergence: FALSE
#>
#>
#> $txvmin
#> Model name: txvmin
#> omega loc scale
#> 136.81 -66.09 18.22
#> Maximum LLF: -166.5841
#> AIC: 339.1683
#> Convergence: TRUE
#>
#>
#> $lxvmin
#> Model name: lxvmin
#> omega loclog scalelog
#> 183.4354 -4.4188 0.4831
#> Maximum LLF: -180.7614
#> AIC: 367.5227
#> Convergence: TRUE
### Draw the graph
mvfplot(fault=sys1g, srms=result)