Skip to content
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

[SPARK-20423][ML] fix MLOR coeffs centering when reg == 0 #17706

Closed
wants to merge 2 commits into from
Closed

[SPARK-20423][ML] fix MLOR coeffs centering when reg == 0 #17706

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1d0fb87
init pr
WeichenXu123 Apr 21, 2017
a694d13
update
WeichenXu123 Apr 21, 2017
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
11 changes: 8 additions & 3 deletions mllib/src/main/scala/org/apache/spark/ml/classification/LogisticRegression.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -609,9 +609,14 @@ class LogisticRegression @Since("1.2.0") (
Friedman, et al. "Regularization Paths for Generalized Linear Models via
Coordinate Descent," https://core.ac.uk/download/files/153/6287975.pdf
*/
val denseValues = denseCoefficientMatrix.values
val coefficientMean = denseValues.sum / denseValues.length
denseCoefficientMatrix.update(_ - coefficientMean)
val centers = Array.fill(numFeatures)(0.0)
denseCoefficientMatrix.foreachActive { case (i, j, v) =>
centers(j) += v
}
centers.transform(_ / numCoefficientSets)
denseCoefficientMatrix.foreachActive { case (i, j, v) =>
denseCoefficientMatrix.update(i, j, v - centers(j))
}
}

// center the intercepts when using multinomial algorithm
Expand Down
6 changes: 6 additions & 0 deletions mllib/src/test/scala/org/apache/spark/ml/classification/LogisticRegressionSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1139,6 +1139,9 @@ class LogisticRegressionSuite
0.10095851, -0.85897154, 0.08392798, 0.07904499), isTransposed = true)
val interceptsR = Vectors.dense(-2.10320093, 0.3394473, 1.76375361)

model1.coefficientMatrix.colIter.foreach(v => assert(v.toArray.sum ~== 0.0 absTol eps))
model2.coefficientMatrix.colIter.foreach(v => assert(v.toArray.sum ~== 0.0 absTol eps))

assert(model1.coefficientMatrix ~== coefficientsR relTol 0.05)
assert(model1.coefficientMatrix.toArray.sum ~== 0.0 absTol eps)
assert(model1.interceptVector ~== interceptsR relTol 0.05)
Expand Down Expand Up @@ -1204,6 +1207,9 @@ class LogisticRegressionSuite
-0.3180040, 0.9679074, -0.2252219, -0.4319914,
0.2452411, -0.6046524, 0.1050710, 0.1180180), isTransposed = true)

model1.coefficientMatrix.colIter.foreach(v => assert(v.toArray.sum ~== 0.0 absTol eps))
Copy link
Contributor

Choose a reason for hiding this comment

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

Before we tested that the coefficients have zero mean using:

assert(model1.coefficientMatrix.toArray.sum ~== 0.0 absTol eps)

We should replace every instance of that test with this new one.

Copy link
Member

Choose a reason for hiding this comment

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

The fix looks right to me. Let's add a test that failing the original implementation.

Copy link
Contributor

Choose a reason for hiding this comment

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

LBFGS seems to automatically find a solution where the coefficients for each feature index sum to zero, so I'm not sure of a way to find a case where this does not happen, TBH.

Copy link
Member

Choose a reason for hiding this comment

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

Interesting. In theory, LBFGS can not see this since all LBFGS knows is the objective function. I think if we feed it with different solutions by justing a constant vector, LBFGS will stop as well. I think maybe it's related to how we setup the initial condition leading to the solution we get.

model2.coefficientMatrix.colIter.foreach(v => assert(v.toArray.sum ~== 0.0 absTol eps))

assert(model1.coefficientMatrix ~== coefficientsR relTol 0.05)
assert(model1.coefficientMatrix.toArray.sum ~== 0.0 absTol eps)
assert(model1.interceptVector.toArray === Array.fill(3)(0.0))
Expand Down