Skip to content

[SPARK-9690][ML][PYTHON] pyspark CrossValidator random seed #10268

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

Closed
jkbradley wants to merge 5 commits into from
Closed

[SPARK-9690][ML][PYTHON] pyspark CrossValidator random seed #10268

Changes from 4 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
8d69a32
SPARK-9690 Adding the possibility to set the seed of the rand in the …
Aug 6, 2015
85aae68
Fixed to use HasSeed
jkbradley Dec 11, 2015
2f60607
fixed import
jkbradley Dec 11, 2015
0bd1457
fixed doc issue
jkbradley Dec 16, 2015
9bf75ad
removed seed from CrossValidator example in tuning.py
jkbradley Dec 16, 2015
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
22 changes: 14 additions & 8 deletions python/pyspark/ml/tuning.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,8 +19,9 @@
import numpy as np

from pyspark import since
from pyspark.ml.param import Params, Param
from pyspark.ml import Estimator, Model
from pyspark.ml.param import Params, Param
from pyspark.ml.param.shared import HasSeed
from pyspark.ml.util import keyword_only
from pyspark.sql.functions import rand

Expand Down Expand Up @@ -89,7 +90,7 @@ def build(self):
return [dict(zip(keys, prod)) for prod in itertools.product(*grid_values)]


class CrossValidator(Estimator):
class CrossValidator(Estimator, HasSeed):
"""
K-fold cross validation.

Expand All @@ -106,7 +107,7 @@ class CrossValidator(Estimator):
>>> lr = LogisticRegression()
>>> grid = ParamGridBuilder().addGrid(lr.maxIter, [0, 1]).build()
>>> evaluator = BinaryClassificationEvaluator()
>>> cv = CrossValidator(estimator=lr, estimatorParamMaps=grid, evaluator=evaluator)
>>> cv = CrossValidator(estimator=lr, estimatorParamMaps=grid, evaluator=evaluator, seed=42)

@yanboliang yanboliang Dec 16, 2015

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

We set default value of seed to this.getClass.getName.hashCode.toLong at Scala side. I think we should keep consistent with that, otherwise users may get different result when training with the same dataset.

@jkbradley jkbradley Dec 16, 2015

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

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

This is just an example though. But I agree we could remove the seed now to simplify the example. I'll do that.

>>> cvModel = cv.fit(dataset)
>>> evaluator.evaluate(cvModel.transform(dataset))
0.8333...
Expand All @@ -129,9 +130,11 @@ class CrossValidator(Estimator):
numFolds = Param(Params._dummy(), "numFolds", "number of folds for cross validation")

@keyword_only
def __init__(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3):
def __init__(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3,
seed=None):
"""
__init__(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3)
__init__(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3,\
seed=None)
"""
super(CrossValidator, self).__init__()
#: param for estimator to be cross-validated
Expand All @@ -151,9 +154,11 @@ def __init__(self, estimator=None, estimatorParamMaps=None, evaluator=None, numF

@keyword_only
@since("1.4.0")
def setParams(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3):
def setParams(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3,
seed=None):
"""
setParams(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3):
setParams(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3,\
seed=None):
Sets params for cross validator.
"""
kwargs = self.setParams._input_kwargs
Expand Down Expand Up @@ -225,9 +230,10 @@ def _fit(self, dataset):
numModels = len(epm)
eva = self.getOrDefault(self.evaluator)
nFolds = self.getOrDefault(self.numFolds)
seed = self.getOrDefault(self.seed)
h = 1.0 / nFolds
randCol = self.uid + "_rand"
df = dataset.select("*", rand(0).alias(randCol))
df = dataset.select("*", rand(seed).alias(randCol))
metrics = np.zeros(numModels)
for i in range(nFolds):
validateLB = i * h
Expand Down