GitHub user BaiGang opened a pull request:
https://github.com/apache/spark/pull/1243
[MLLIB] SPARK-2303: Poisson regression model for count data
This pull request includes the implementations of Poisson regression in
mllib.regression for modeling count data. In detail, it includes:
1. The gradient of the negative log-likelihood of Poisson regression model.
2. The implementations of PoissonRegressionModel, including the
generalized linear algorithm class which uses L-BFGS and SGD for parameter
estimation respectively and the companion objects.
3. The test suites
* the gradient/loss computation
* the regression method using LBFGS optimization on generated data set
* the regression method using LBFGS optimization on real-world data set
* the regression method using SGD optimization on generated data set
* the regression method using SGD optimization on real-world data set
4. a Poisson regression data generator in mllib/util for producing the
test data.
JIRA: https://issues.apache.org/jira/browse/SPARK-2303
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/BaiGang/spark poisson
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/spark/pull/1243.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #1243
----
commit abf543d3f36a02e5dbbad797ff8f84c043855469
Author: Gang Bai <[email protected]>
Date: 2014-06-27T05:35:24Z
The implementations of Poission regression in mllib/regression. It includes
1)the gradient of the negative log-likelihood, 2)the implementation of
PoissonRegressionModel, the generalized linear algorithm class which uses
L-BFGS and SGD for parameter estimation respectively, 3) the test suites for
the gradient/loss computation, the regression method on generated and
real-world data set, and 4) a Poisson regression data generator in mllib/util
for producing the test data.
----
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
