Estimates the runtimes of jobs using the random forest implemented in ranger.
Observed runtimes are retrieved from the Registry and runtimes are
predicted for unfinished jobs.
The estimated remaining time is calculated in the print method.
You may also pass n here to determine the number of parallel jobs which is then used
in a simple Longest Processing Time (LPT) algorithm to give an estimate for the parallel runtime.
estimateRuntimes(tab, ..., reg = getDefaultRegistry())
# S3 method for class 'RuntimeEstimate'
print(x, n = 1L, ...)[data.table]
Table with column “job.id” and additional columns to predict the runtime.
Observed runtimes will be looked up in the registry and serve as dependent variable.
All columns in tab except “job.id” will be passed to ranger as
independent variables to fit the model.
[ANY]
Additional parameters passed to ranger. Ignored for the print method.
[Registry]
Registry. If not explicitly passed, uses the default registry (see setDefaultRegistry).
[RuntimeEstimate]
Object to print.
[integer(1)]
Number of parallel jobs to assume for runtime estimation.
[RuntimeEstimate] which is a list with two named elements:
“runtimes” is a data.table with columns “job.id”,
“runtime” (in seconds) and “type” (“estimated” if runtime is estimated,
“observed” if runtime was observed).
The other element of the list named “model”] contains the fitted random forest object.
# Create a simple toy registry
set.seed(1)
tmp = makeExperimentRegistry(file.dir = NA, make.default = FALSE, seed = 1)
#> No readable configuration file found
#> Created registry in '/tmp/batchtools-example/reg' using cluster functions 'Interactive'
addProblem(name = "iris", data = iris, fun = function(data, ...) nrow(data), reg = tmp)
#> Adding problem 'iris'
addAlgorithm(name = "nrow", function(instance, ...) nrow(instance), reg = tmp)
#> Adding algorithm 'nrow'
addAlgorithm(name = "ncol", function(instance, ...) ncol(instance), reg = tmp)
#> Adding algorithm 'ncol'
addExperiments(algo.designs = list(nrow = data.table::CJ(x = 1:50, y = letters[1:5])), reg = tmp)
#> Adding 250 experiments ('iris'[1] x 'nrow'[250] x repls[1]) ...
addExperiments(algo.designs = list(ncol = data.table::CJ(x = 1:50, y = letters[1:5])), reg = tmp)
#> Adding 250 experiments ('iris'[1] x 'ncol'[250] x repls[1]) ...
# We use the job parameters to predict runtimes
tab = unwrap(getJobPars(reg = tmp))
# First we need to submit some jobs so that the forest can train on some data.
# Thus, we just sample some jobs from the registry while grouping by factor variables.
library(data.table)
ids = tab[, .SD[sample(nrow(.SD), 5)], by = c("problem", "algorithm", "y")]
setkeyv(ids, "job.id")
submitJobs(ids, reg = tmp)
#> Submitting 50 jobs in 50 chunks using cluster functions 'Interactive' ...
waitForJobs(reg = tmp)
#> [1] TRUE
# We "simulate" some more realistic runtimes here to demonstrate the functionality:
# - Algorithm "ncol" is 5 times more expensive than "nrow"
# - x has no effect on the runtime
# - If y is "a" or "b", the runtimes are really high
runtime = function(algorithm, x, y) {
ifelse(algorithm == "nrow", 100L, 500L) + 1000L * (y %in% letters[1:2])
}
tmp$status[ids, done := done + tab[ids, runtime(algorithm, x, y)]]
#> Key: <job.id>
#> job.id def.id submitted started done error mem.used resource.id
#> <int> <int> <num> <num> <num> <char> <num> <int>
#> 1: 1 1 NA NA NA <NA> NA NA
#> 2: 2 2 NA NA NA <NA> NA NA
#> 3: 3 3 NA NA NA <NA> NA NA
#> 4: 4 4 NA NA NA <NA> NA NA
#> 5: 5 5 NA NA NA <NA> NA NA
#> ---
#> 496: 496 496 NA NA NA <NA> NA NA
#> 497: 497 497 NA NA NA <NA> NA NA
#> 498: 498 498 NA NA NA <NA> NA NA
#> 499: 499 499 1760453475 1760453475 1760453975 <NA> NA 1
#> 500: 500 500 NA NA NA <NA> NA NA
#> batch.id log.file job.hash job.name repl
#> <char> <char> <char> <char> <int>
#> 1: <NA> <NA> <NA> <NA> 1
#> 2: <NA> <NA> <NA> <NA> 1
#> 3: <NA> <NA> <NA> <NA> 1
#> 4: <NA> <NA> <NA> <NA> 1
#> 5: <NA> <NA> <NA> <NA> 1
#> ---
#> 496: <NA> <NA> <NA> <NA> 1
#> 497: <NA> <NA> <NA> <NA> 1
#> 498: <NA> <NA> <NA> <NA> 1
#> 499: cfInteractive <NA> jobd3307f1c7044bae0d9b5eaaa4cb90db8 <NA> 1
#> 500: <NA> <NA> <NA> <NA> 1
rjoin(sjoin(tab, ids), getJobStatus(ids, reg = tmp)[, c("job.id", "time.running")])
#> Key: <job.id>
#> job.id problem algorithm x y time.running
#> <int> <char> <char> <int> <char> <difftime>
#> 1: 32 iris nrow 7 b 1100.0533 secs
#> 2: 42 iris nrow 9 b 1100.0317 secs
#> 3: 47 iris nrow 10 b 1100.0353 secs
#> 4: 66 iris nrow 14 a 1100.0412 secs
#> 5: 73 iris nrow 15 c 100.0343 secs
#> 6: 75 iris nrow 15 e 100.0535 secs
#> 7: 86 iris nrow 18 a 1100.0426 secs
#> 8: 100 iris nrow 20 e 100.0373 secs
#> 9: 101 iris nrow 21 a 1100.0332 secs
#> 10: 103 iris nrow 21 c 100.0301 secs
#> 11: 123 iris nrow 25 c 100.0434 secs
#> 12: 125 iris nrow 25 e 100.0345 secs
#> 13: 161 iris nrow 33 a 1100.0400 secs
#> 14: 165 iris nrow 33 e 100.0383 secs
#> 15: 169 iris nrow 34 d 100.0433 secs
#> 16: 183 iris nrow 37 c 100.0352 secs
#> 17: 184 iris nrow 37 d 100.0346 secs
#> 18: 203 iris nrow 41 c 100.0341 secs
#> 19: 207 iris nrow 42 b 1100.0466 secs
#> 20: 209 iris nrow 42 d 100.0372 secs
#> 21: 220 iris nrow 44 e 100.0535 secs
#> 22: 227 iris nrow 46 b 1100.0371 secs
#> 23: 229 iris nrow 46 d 100.0347 secs
#> 24: 231 iris nrow 47 a 1100.0365 secs
#> 25: 244 iris nrow 49 d 100.0466 secs
#> 26: 260 iris ncol 2 e 500.0564 secs
#> 27: 276 iris ncol 6 a 1500.0487 secs
#> 28: 278 iris ncol 6 c 500.0709 secs
#> 29: 279 iris ncol 6 d 500.0404 secs
#> 30: 296 iris ncol 10 a 1500.0523 secs
#> 31: 320 iris ncol 14 e 500.0431 secs
#> 32: 340 iris ncol 18 e 500.0332 secs
#> 33: 347 iris ncol 20 b 1500.0318 secs
#> 34: 363 iris ncol 23 c 500.0442 secs
#> 35: 369 iris ncol 24 d 500.0553 secs
#> 36: 373 iris ncol 25 c 500.0509 secs
#> 37: 387 iris ncol 28 b 1500.0392 secs
#> 38: 410 iris ncol 32 e 500.0345 secs
#> 39: 421 iris ncol 35 a 1500.0363 secs
#> 40: 436 iris ncol 38 a 1500.0327 secs
#> 41: 444 iris ncol 39 d 500.0334 secs
#> 42: 448 iris ncol 40 c 500.0333 secs
#> 43: 456 iris ncol 42 a 1500.0321 secs
#> 44: 459 iris ncol 42 d 500.0435 secs
#> 45: 467 iris ncol 44 b 1500.0350 secs
#> 46: 468 iris ncol 44 c 500.0366 secs
#> 47: 475 iris ncol 45 e 500.0355 secs
#> 48: 482 iris ncol 47 b 1500.0350 secs
#> 49: 492 iris ncol 49 b 1500.0534 secs
#> 50: 499 iris ncol 50 d 500.0457 secs
#> job.id problem algorithm x y time.running
# Estimate runtimes:
est = estimateRuntimes(tab, reg = tmp)
print(est)
#> Runtime Estimate for 500 jobs with 1 CPUs
#> Done : 0d 09h 43m 22.0s
#> Remaining: 3d 17h 37m 49.3s
#> Total : 4d 03h 21m 11.4s
rjoin(tab, est$runtimes)
#> Key: <job.id>
#> job.id problem algorithm x y type runtime
#> <int> <char> <char> <int> <char> <fctr> <num>
#> 1: 1 iris nrow 1 a estimated 1107.2567
#> 2: 2 iris nrow 1 b estimated 1091.2113
#> 3: 3 iris nrow 1 c estimated 338.3696
#> 4: 4 iris nrow 1 d estimated 318.6494
#> 5: 5 iris nrow 1 e estimated 318.6948
#> ---
#> 496: 496 iris ncol 50 a estimated 1383.8348
#> 497: 497 iris ncol 50 b estimated 1391.0858
#> 498: 498 iris ncol 50 c estimated 619.0990
#> 499: 499 iris ncol 50 d observed 500.0457
#> 500: 500 iris ncol 50 e estimated 580.0901
print(est, n = 10)
#> Runtime Estimate for 500 jobs with 10 CPUs
#> Done : 0d 09h 43m 22.0s
#> Remaining: 3d 17h 37m 49.3s
#> Parallel : 0d 08h 58m 29.1s
#> Total : 4d 03h 21m 11.4s
# Submit jobs with longest runtime first:
ids = est$runtimes[type == "estimated"][order(runtime, decreasing = TRUE)]
print(ids)
#> job.id type runtime
#> <int> <fctr> <num>
#> 1: 466 estimated 1421.1282
#> 2: 461 estimated 1419.7348
#> 3: 462 estimated 1415.9087
#> 4: 457 estimated 1415.9086
#> 5: 472 estimated 1415.0621
#> ---
#> 446: 194 estimated 132.4721
#> 447: 189 estimated 131.8054
#> 448: 204 estimated 131.7223
#> 449: 174 estimated 131.1890
#> 450: 179 estimated 130.0412
if (FALSE) { # \dontrun{
submitJobs(ids, reg = tmp)
} # }
# Group jobs into chunks with runtime < 1h
ids = est$runtimes[type == "estimated"]
ids[, chunk := binpack(runtime, 3600)]
#> Key: <job.id>
#> job.id type runtime chunk
#> <int> <fctr> <num> <int>
#> 1: 1 estimated 1107.2567 47
#> 2: 2 estimated 1091.2113 51
#> 3: 3 estimated 338.3696 37
#> 4: 4 estimated 318.6494 70
#> 5: 5 estimated 318.6948 33
#> ---
#> 446: 495 estimated 587.0244 14
#> 447: 496 estimated 1383.8348 19
#> 448: 497 estimated 1391.0858 13
#> 449: 498 estimated 619.0990 4
#> 450: 500 estimated 580.0901 17
print(ids)
#> Key: <job.id>
#> job.id type runtime chunk
#> <int> <fctr> <num> <int>
#> 1: 1 estimated 1107.2567 47
#> 2: 2 estimated 1091.2113 51
#> 3: 3 estimated 338.3696 37
#> 4: 4 estimated 318.6494 70
#> 5: 5 estimated 318.6948 33
#> ---
#> 446: 495 estimated 587.0244 14
#> 447: 496 estimated 1383.8348 19
#> 448: 497 estimated 1391.0858 13
#> 449: 498 estimated 619.0990 4
#> 450: 500 estimated 580.0901 17
print(ids[, list(runtime = sum(runtime)), by = chunk])
#> chunk runtime
#> <int> <num>
#> 1: 47 3493.903
#> 2: 51 3595.348
#> 3: 37 3597.163
#> 4: 70 3490.491
#> 5: 33 3599.662
#> 6: 54 3589.077
#> 7: 71 3488.381
#> 8: 48 3492.394
#> 9: 52 3599.825
#> 10: 55 3580.492
#> 11: 72 3485.419
#> 12: 68 3495.168
#> 13: 56 3566.849
#> 14: 73 3478.614
#> 15: 69 3493.178
#> 16: 46 3519.658
#> 17: 50 3599.956
#> 18: 38 3598.919
#> 19: 64 3515.549
#> 20: 34 3599.756
#> 21: 39 3596.919
#> 22: 65 3512.143
#> 23: 62 3519.006
#> 24: 43 3579.570
#> 25: 61 3529.298
#> 26: 60 3533.421
#> 27: 53 3599.578
#> 28: 40 3598.912
#> 29: 58 3547.824
#> 30: 57 3560.195
#> 31: 49 3484.275
#> 32: 42 3586.102
#> 33: 59 3537.205
#> 34: 35 3599.877
#> 35: 41 3587.732
#> 36: 36 3598.976
#> 37: 44 3547.085
#> 38: 63 3517.317
#> 39: 66 3506.777
#> 40: 45 3545.898
#> 41: 67 3505.802
#> 42: 28 3593.112
#> 43: 24 3596.859
#> 44: 26 3587.720
#> 45: 25 3598.983
#> 46: 29 3567.741
#> 47: 23 3599.829
#> 48: 75 3599.847
#> 49: 20 3520.988
#> 50: 74 3472.919
#> 51: 9 3590.197
#> 52: 21 3514.818
#> 53: 12 3562.487
#> 54: 8 3594.973
#> 55: 6 3591.969
#> 56: 5 3593.786
#> 57: 10 3577.467
#> 58: 32 3599.440
#> 59: 11 3575.203
#> 60: 81 3480.912
#> 61: 83 3599.907
#> 62: 76 3593.037
#> 63: 3 3599.636
#> 64: 79 3501.757
#> 65: 80 3492.301
#> 66: 86 3580.240
#> 67: 88 3560.468
#> 68: 91 2422.115
#> 69: 82 3470.248
#> 70: 78 3511.647
#> 71: 89 3551.825
#> 72: 90 3528.456
#> 73: 2 3599.707
#> 74: 77 3526.878
#> 75: 85 3587.614
#> 76: 87 3569.977
#> 77: 84 3593.995
#> 78: 4 3599.018
#> 79: 7 3596.572
#> 80: 30 3563.571
#> 81: 18 3575.196
#> 82: 14 3596.808
#> 83: 22 3599.506
#> 84: 19 3571.153
#> 85: 17 3583.873
#> 86: 31 3550.507
#> 87: 13 3598.899
#> 88: 27 3583.829
#> 89: 15 3599.764
#> 90: 16 3597.165
#> 91: 1 3470.720
#> chunk runtime
if (FALSE) { # \dontrun{
submitJobs(ids, reg = tmp)
} # }
# Group jobs into 10 chunks with similar runtime
ids = est$runtimes[type == "estimated"]
ids[, chunk := lpt(runtime, 10)]
#> Key: <job.id>
#> job.id type runtime chunk
#> <int> <fctr> <num> <int>
#> 1: 1 estimated 1107.2567 3
#> 2: 2 estimated 1091.2113 10
#> 3: 3 estimated 338.3696 6
#> 4: 4 estimated 318.6494 7
#> 5: 5 estimated 318.6948 5
#> ---
#> 446: 495 estimated 587.0244 9
#> 447: 496 estimated 1383.8348 9
#> 448: 497 estimated 1391.0858 3
#> 449: 498 estimated 619.0990 2
#> 450: 500 estimated 580.0901 2
print(ids[, list(runtime = sum(runtime)), by = chunk])
#> chunk runtime
#> <int> <num>
#> 1: 3 32231.74
#> 2: 10 32308.50
#> 3: 6 32230.86
#> 4: 7 32308.83
#> 5: 5 32234.56
#> 6: 4 32230.05
#> 7: 1 32309.06
#> 8: 8 32231.39
#> 9: 2 32292.48
#> 10: 9 32291.88