Skip to content

liquidSVM #2428

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
larskotthoff merged 6 commits into from
Sep 24, 2018
Merged

liquidSVM #2428

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
c13335a
create liquidSVM learners
PhilippPro Aug 31, 2018
8688547
add the necessary package
PhilippPro Sep 6, 2018
15b2d0c
include some tests (that are still failing (probably))
PhilippPro Sep 7, 2018
6ceb849
add debug seed
PhilippPro Sep 11, 2018
e6feccb
change parameter settings of liquidSVM
PhilippPro Sep 24, 2018
013b9cf
Merge branch 'master' into liquidSVM
larskotthoff Sep 24, 2018
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
include some tests (that are still failing (probably))
  • Loading branch information
@PhilippPro
PhilippPro committed Sep 7, 2018
commit 15b2d0cf060129a50c8c97cd48d1c3dc15fe7981
20 changes: 10 additions & 10 deletions R/RLearner_classif_liquidSVM.R
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,24 +5,24 @@ makeRLearner.classif.liquidSVM = function() {
package = "liquidSVM",
par.set = makeParamSet(
makeIntegerLearnerParam(id = "d", lower = 0L),
Copy link
Member

@larskotthoff larskotthoff Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This should not be tunable (tunable = FALSE).

makeLogicalLearnerParam(id = "scale", default = TRUE),
makeIntegerLearnerParam(id = "threads", default = 1L, lower = -1L),
makeIntegerLearnerParam(id = "partition_choice", default = 0L, lower = 0L),
makeIntegerLearnerParam(id = "grid_choice", default = 0L, lower = -2L, upper = 2L),
makeIntegerLearnerParam(id = "adaptivity_control", default = 0L, lower = 0L, upper = 2L),
makeIntegerLearnerParam(id = "random_seed", default = 1L),
makeLogicalLearnerParam(id = "scale"),
Copy link
Member

@larskotthoff larskotthoff Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Default for this is TRUE.

makeIntegerLearnerParam(id = "threads", lower = -1L),
Copy link
Member

@larskotthoff larskotthoff Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Default is 0.

makeIntegerLearnerParam(id = "partition_choice", lower = 0L),
Copy link
Member

@larskotthoff larskotthoff Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Default is 0.

makeIntegerLearnerParam(id = "grid_choice", lower = -2L, upper = 2L),
Copy link
Member

@larskotthoff larskotthoff Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

How does this interact with gamma_steps, lambda_steps, etc? Are these options mutually exclusive?

Copy link
Member

@larskotthoff larskotthoff Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

And all of these options should be not tunable.

makeIntegerLearnerParam(id = "adaptivity_control", lower = 0L, upper = 2L),
Copy link
Member

@larskotthoff larskotthoff Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Default is 0.

makeIntegerLearnerParam(id = "random_seed", default = 1),
Copy link
Member

@larskotthoff larskotthoff Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I didn't find any documentation on what the default random seed is -- where did you find this? It also shouldn't be tunable.

makeIntegerLearnerParam(id = "fold"),
Copy link
Member

@larskotthoff larskotthoff Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should that be folds? And there should be a minimum for this (and again not tunable).

makeIntegerLearnerParam(id = "clipping"),
Copy link
Member

@larskotthoff larskotthoff Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seems that the minimum is -1. Also probably shouldn't be tunable.

makeIntegerLearnerParam(id = "gamma_steps", lower = 0),
makeNumericLearnerParam(id = "min_gamma", lower = 0),
makeNumericLearnerParam(id = "max_gamma", lower = 0),
makeNumericLearnerParam(id = "gamma", lower = 0),
makeNumericVectorLearnerParam(id = "gammas", lower = 0),
makeIntegerLearnerParam(id = "lambda_steps", lower = 0),
makeNumericLearnerParam(id = "min_lambda", lower = 0),
makeNumericLearnerParam(id = "max_lambda", lower = 0),
makeNumericLearnerParam(id = "lambdas", lower = 0),
makeNumericLearnerParam(id = "c_values", lower = 0),
makeLogicalLearnerParam(id = "useCells", default = FALSE)
makeNumericVectorLearnerParam(id = "lambdas", lower = 0),
makeNumericVectorLearnerParam(id = "c_values", lower = 0),
makeLogicalLearnerParam(id = "useCells")
Copy link
Member

@larskotthoff larskotthoff Sep 10, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seems that the default for this is FALSE. Should be not tunable.

),
properties = c("twoclass", "multiclass", "numerics", "factors"),
name = "Support Vector Machines (liquidSVM)",
Expand Down
20 changes: 10 additions & 10 deletions R/RLearner_regr_liquidSVM.R
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,24 +5,24 @@ makeRLearner.regr.liquidSVM = function() {
package = "liquidSVM",
par.set = makeParamSet(
makeIntegerLearnerParam(id = "d", lower = 0L),
makeLogicalLearnerParam(id = "scale", default = TRUE),
makeIntegerLearnerParam(id = "threads", default = 1L, lower = -1L),
makeIntegerLearnerParam(id = "partition_choice", default = 0L, lower = 0L),
makeIntegerLearnerParam(id = "grid_choice", default = 0L, lower = -2L, upper = 2L),
makeIntegerLearnerParam(id = "adaptivity_control", default = 0L, lower = 0L, upper = 2L),
makeIntegerLearnerParam(id = "random_seed", default = 1L),
makeLogicalLearnerParam(id = "scale"),
makeIntegerLearnerParam(id = "threads", lower = -1L),
makeIntegerLearnerParam(id = "partition_choice", lower = 0L),
makeIntegerLearnerParam(id = "grid_choice", lower = -2L, upper = 2L),
makeIntegerLearnerParam(id = "adaptivity_control", lower = 0L, upper = 2L),
makeIntegerLearnerParam(id = "random_seed", default = 1),
makeIntegerLearnerParam(id = "fold"),
makeIntegerLearnerParam(id = "clipping"),
makeIntegerLearnerParam(id = "gamma_steps", lower = 0),
makeNumericLearnerParam(id = "min_gamma", lower = 0),
makeNumericLearnerParam(id = "max_gamma", lower = 0),
makeNumericLearnerParam(id = "gamma", lower = 0),
makeNumericVectorLearnerParam(id = "gammas", lower = 0),
makeIntegerLearnerParam(id = "lambda_steps", lower = 0),
makeNumericLearnerParam(id = "min_lambda", lower = 0),
makeNumericLearnerParam(id = "max_lambda", lower = 0),
makeNumericLearnerParam(id = "lambdas", lower = 0),
makeNumericLearnerParam(id = "c_values", lower = 0),
makeLogicalLearnerParam(id = "useCells", default = FALSE)
makeNumericVectorLearnerParam(id = "lambdas", lower = 0),
makeNumericVectorLearnerParam(id = "c_values", lower = 0),
makeLogicalLearnerParam(id = "useCells")
),
properties = c( "numerics", "factors"),
name = "Support Vector Machines (liquidSVM)",
Expand Down
38 changes: 38 additions & 0 deletions tests/testthat/test_classif_liquidSVM.R
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
context("classif_liquidSVM")

test_that("classif_liquidSVM", {
requirePackagesOrSkip("liquidSVM", default.method = "load")
parset.list = list(
list(),
list(partition_choice = 6),
list(partition_choice = 5),
list(grid_choice = 1),
list(grid_choice = 2),
list(adaptivity_control = 1),
list(adaptivity_control = 2),
list(clipping = -1),
list(clipping = 0),
list(gamma_steps = 5, min_gamma = 0.1, max_gamma = 25, lambda_steps = 5, min_lambda = 0.1, max_lambda = 25),
list(useCells = TRUE),
list(gammas = c(0.1,1,10), lambdas = c(0.1,1,10), c_values = c(0.1,1,10))
)
# Kernel, more advanced parameters...

old.predicts.list = list()
for (i in seq_along(parset.list)) {
parset = parset.list[[i]]
pars = list(x = multiclass.formula, y = multiclass.train)
pars = c(pars, parset)
set.seed(getOption("mlr.debug.seed"))
set.seed(123)
m = do.call(liquidSVM::svm, pars)
set.seed(getOption("mlr.debug.seed"))
set.seed(123)
p = predict(m, newdata = multiclass.test, type = "response")
old.predicts.list[[i]] = p
}

testSimpleParsets("classif.liquidSVM", multiclass.df, multiclass.target,
multiclass.train.inds, old.predicts.list, parset.list[1])

})
33 changes: 33 additions & 0 deletions tests/testthat/test_regr_liquidSVM.R
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,33 @@
context("regr_liquidSVM")

test_that("regr_liquidSVM", {
requirePackagesOrSkip("liquidSVM", default.method = "load")

parset.list = list(
list(),
list(partition_choice = 6),
list(partition_choice = 5),
list(grid_choice = 1),
list(grid_choice = 2),
list(adaptivity_control = 1),
list(adaptivity_control = 2),
list(clipping = -1),
list(clipping = 0),
list(gamma_steps = 5, min_gamma = 0.1, max_gamma = 25, lambda_steps = 5, min_lambda = 0.1, max_lambda = 25),
list(useCells = TRUE),
list(gammas = c(0.1,1,10), lambdas = c(0.1,1,10), c_values = c(0.1,1,10))
)
old.predicts.list = list()
for (i in seq_along(parset.list)) {
parset = parset.list[[i]]
pars = list(x = regr.formula, y = regr.train)
pars = c(pars, parset)
set.seed(getOption("mlr.debug.seed"))
m = do.call(liquidSVM::svm, pars)
p = predict(m, newdata = regr.test)
old.predicts.list[[i]] = p
}

testSimpleParsets("regr.liquidSVM", regr.df, regr.target, regr.train.inds, old.predicts.list, parset.list)
})