...
No. | Item | Description | Link | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | Improved error message for Elastic Net predict() | When we pass the selected coefficients to elastic net's "predict()" function, it throws as ugly error message which is not indicative of the real error. |
| ||||||||||
2 | Confusing Error Messages while running elastic net prediction function | Fix confusing error message |
| ||||||||||
3 | LDA (parsed) model table and output table disagree | Investigate and determine if this is an issue. If it is, repair it. |
| ||||||||||
4 | PivotalR test failures indicate potential bugs in MADlib GLM | These problems may be just numerical issues with too large the condition numbers or too small of a training set. To be investigated. |
| ||||||||||
5 | Implement skipping of arrays-with-NULL for elastic net predict | Better NULL handling for elastic net predict. |
| ||||||||||
6 | Improve RF output format for variable importance | Easier way of accessing the variable importance output from random forest so that I can understand which are the most important variables. |
| ||||||||||
7 | Covariance matrix | Add parameter to output covariance matrix to Pierson's correlation function. |
| ||||||||||
8 |
New
...
No. | Item | Description | Link | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
1 | Add PMML export modules* | Support additional MADlib modules for PMML export |
| ||||||||
2 |
*Some notes on PMML below...
MADlib models can be exported in PMML format for use in scoring by a PMML evaluator.
The following MADlib 1.9 algorithms can be exported in PMML format:
Linear regression
Logistic regression
GLM
Multinomial regression
Ordinal regression
Decision trees
Random forest
Your contribution here...
The Predictive Model Markup Language (PMML) is an XML-based file format that provides a way for applications to describe and exchange models produced by data mining and machine learning algorithms.
For more information, please see http://www.dmg.org/
JPMML is an open source PMML evaluator available under GPL license.
For more information, please see https://github.com/jpmml and https://github.com/jpmml/openscoring
You can only export from MADlib into PMML (no import currently)
New Non-Iterative Modules
No. | Item | Description | Link | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | k-Nearest Neighbors | Initial implementation of k-NN |
| ||||||||||
2 | Stratified sampling | Utility to perform stratified, randomized, proportional sampling and labeling. |
| ||||||||||
3 | URI utilities | A set of utilities for parsing and extracting URIs from text. |
| ||||||||||
4 | Anonymization | Utility for anonymization. |
| ||||||||||
5 | Sessionization | Utility to partition event streams into sessions by timeouts and identifiers |
| ||||||||||
6 | Mixed Effects Modeling | Mixed-effects model containing fixed-effects and random-effects components. |
| ||||||||||
7 | New Methods: Expectation Maximimization | Gaussian mixture modeling and others |
| ||||||||||
8 |
...