[SPARK-3066][MLLIB] Support recommendAll in matrix factorization model #5829
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…pute map measure along with rmse
…tric for movielens dataset
… BoundedPriorityQueue similar to RDD.top
…for product recommendation per user using randomized split
…batch predict APIs in matrix factorization
|
@mengxr looks good to me...I will fix SPARK-4321 based on this merge...I need blockify for rowSimilarities (tall skinny sparse matrices for row similarities)...should we extract it out to IndexedRow ? I can do that cleanup in my row similarities PR... |
|
Test build #31524 has finished for PR 5829 at commit
|
There was a problem hiding this comment.
I have to look closely into it tomorrow...I have been using topByKey internally and did not remember seeing this bug...
There was a problem hiding this comment.
yup topByKey behavior as implemented was correct...
Contributor
Author
There was a problem hiding this comment.
It was correct. Pushed an update.
|
Test build #31525 has finished for PR 5829 at commit
|
|
Test build #31530 has finished for PR 5829 at commit
|
There was a problem hiding this comment.
Normally items are skinny ~ 1M...and ranks are low...50...so 1Mx50 bytes ~ 50 MB...with 8M products, its 400 MB...I still think that cartesian will be slower than the version I added in terms of runtime....did you run any benchmark with the old code ?
Contributor
Author
There was a problem hiding this comment.
That depends on the data. It is also common to have near-squared rating matrix. This should provide similar performance if the items/products are not super small, but I didn't test the performance. The advantage is that this approach doesn't touch the driver, so it could be more scalable.
There was a problem hiding this comment.
I also like it better as it should scale fine assuming cartesian keys are under control...say to 100M x 10M with 400 factors....
|
Test build #31541 has finished for PR 5829 at commit
|
Contributor
Author
|
Okay, I'm merging this in. Please submit a PR for SPARK-4231 if you get time to work on it. Thanks! |
jeanlyn
pushed a commit
to jeanlyn/spark
that referenced
this pull request
May 28, 2015
This is based on apache#3098 from debasish83. 1. BLAS' GEMM is used to compute inner products. 2. Reverted changes to MovieLensALS. SPARK-4231 should be addressed in a separate PR. 3. ~~Fixed a bug in topByKey~~ Closes apache#3098 debasish83 coderxiang Author: Debasish Das <debasish.das@one.verizon.com> Author: Xiangrui Meng <meng@databricks.com> Closes apache#5829 from mengxr/SPARK-3066 and squashes the following commits: 22e6a87 [Xiangrui Meng] topByKey was correct. update its usage 389b381 [Xiangrui Meng] fix indentation 49953de [Xiangrui Meng] Merge remote-tracking branch 'apache/master' into SPARK-3066 cb9799a [Xiangrui Meng] revert MovieLensALS f864f5e [Xiangrui Meng] update test and fix a bug in topByKey c5e0181 [Xiangrui Meng] use GEMM and topByKey 3a0c4eb [Debasish Das] updated with spark master 98fa424 [Debasish Das] updated with master ee99571 [Debasish Das] addressed initial review comments;merged with master;added tests for batch predict APIs in matrix factorization 3f97c49 [Debasish Das] fixed spark coding style for imports 7163a5c [Debasish Das] Added API for batch user and product recommendation; MAP calculation for product recommendation per user using randomized split d144f57 [Debasish Das] recommendAll API to MatrixFactorizationModel, uses topK finding using BoundedPriorityQueue similar to RDD.top f38a1b5 [Debasish Das] use sampleByKey for per user sampling 10cbb37 [Debasish Das] provide ratio for topN product validation; generate MAP and prec@k metric for movielens dataset 9fa063e [Debasish Das] import scala.math.round 4bbae0f [Debasish Das] comments fixed as per scalastyle cd3ab31 [Debasish Das] merged with AbstractParams serialization bug 9b3951f [Debasish Das] validate user/product on MovieLens dataset through user input and compute map measure along with rmse
jeanlyn
pushed a commit
to jeanlyn/spark
that referenced
this pull request
Jun 12, 2015
This is based on apache#3098 from debasish83. 1. BLAS' GEMM is used to compute inner products. 2. Reverted changes to MovieLensALS. SPARK-4231 should be addressed in a separate PR. 3. ~~Fixed a bug in topByKey~~ Closes apache#3098 debasish83 coderxiang Author: Debasish Das <debasish.das@one.verizon.com> Author: Xiangrui Meng <meng@databricks.com> Closes apache#5829 from mengxr/SPARK-3066 and squashes the following commits: 22e6a87 [Xiangrui Meng] topByKey was correct. update its usage 389b381 [Xiangrui Meng] fix indentation 49953de [Xiangrui Meng] Merge remote-tracking branch 'apache/master' into SPARK-3066 cb9799a [Xiangrui Meng] revert MovieLensALS f864f5e [Xiangrui Meng] update test and fix a bug in topByKey c5e0181 [Xiangrui Meng] use GEMM and topByKey 3a0c4eb [Debasish Das] updated with spark master 98fa424 [Debasish Das] updated with master ee99571 [Debasish Das] addressed initial review comments;merged with master;added tests for batch predict APIs in matrix factorization 3f97c49 [Debasish Das] fixed spark coding style for imports 7163a5c [Debasish Das] Added API for batch user and product recommendation; MAP calculation for product recommendation per user using randomized split d144f57 [Debasish Das] recommendAll API to MatrixFactorizationModel, uses topK finding using BoundedPriorityQueue similar to RDD.top f38a1b5 [Debasish Das] use sampleByKey for per user sampling 10cbb37 [Debasish Das] provide ratio for topN product validation; generate MAP and prec@k metric for movielens dataset 9fa063e [Debasish Das] import scala.math.round 4bbae0f [Debasish Das] comments fixed as per scalastyle cd3ab31 [Debasish Das] merged with AbstractParams serialization bug 9b3951f [Debasish Das] validate user/product on MovieLens dataset through user input and compute map measure along with rmse
nemccarthy
pushed a commit
to nemccarthy/spark
that referenced
this pull request
Jun 19, 2015
This is based on apache#3098 from debasish83. 1. BLAS' GEMM is used to compute inner products. 2. Reverted changes to MovieLensALS. SPARK-4231 should be addressed in a separate PR. 3. ~~Fixed a bug in topByKey~~ Closes apache#3098 debasish83 coderxiang Author: Debasish Das <debasish.das@one.verizon.com> Author: Xiangrui Meng <meng@databricks.com> Closes apache#5829 from mengxr/SPARK-3066 and squashes the following commits: 22e6a87 [Xiangrui Meng] topByKey was correct. update its usage 389b381 [Xiangrui Meng] fix indentation 49953de [Xiangrui Meng] Merge remote-tracking branch 'apache/master' into SPARK-3066 cb9799a [Xiangrui Meng] revert MovieLensALS f864f5e [Xiangrui Meng] update test and fix a bug in topByKey c5e0181 [Xiangrui Meng] use GEMM and topByKey 3a0c4eb [Debasish Das] updated with spark master 98fa424 [Debasish Das] updated with master ee99571 [Debasish Das] addressed initial review comments;merged with master;added tests for batch predict APIs in matrix factorization 3f97c49 [Debasish Das] fixed spark coding style for imports 7163a5c [Debasish Das] Added API for batch user and product recommendation; MAP calculation for product recommendation per user using randomized split d144f57 [Debasish Das] recommendAll API to MatrixFactorizationModel, uses topK finding using BoundedPriorityQueue similar to RDD.top f38a1b5 [Debasish Das] use sampleByKey for per user sampling 10cbb37 [Debasish Das] provide ratio for topN product validation; generate MAP and prec@k metric for movielens dataset 9fa063e [Debasish Das] import scala.math.round 4bbae0f [Debasish Das] comments fixed as per scalastyle cd3ab31 [Debasish Das] merged with AbstractParams serialization bug 9b3951f [Debasish Das] validate user/product on MovieLens dataset through user input and compute map measure along with rmse
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is based on #3098 from @debasish83.
Fixed a bug in topByKeyCloses #3098
@debasish83 @coderxiang