Implements generalized linear multitask learning and solves multiple related sparse linear regression problems simultaneously. This action is used for GRAPHMULTIREG operations.
| Parameter | Description |
|---|---|
| attributes | Specifies variable attributes. Refer to common `casinvardesc` for full details (Appendix A: Common Parameters). |
| graphOut | Specifies the output data table to save the graph table. Refer to common `casouttable (Form 1)` for full details (Appendix A: Common Parameters). |
| graphTable | Specifies the graph table. Its first column must be the row ID. Must include all specified target variables. Refer to common `castable (Form 1)` for full details (Appendix A: Common Parameters). |
| graphType | Specifies the type of graph table. Default: CLUSTER. Options: CLUSTER (all tasks connected), CUSTOM (user-specified), FUSE (each task connected to the next), INDEP (all tasks independent). |
| inputs | Specifies variables for training. Refer to common `casinvardesc` for full details (Appendix A: Common Parameters). |
| maxIters | Specifies the maximum number of iterations. Default: 100. |
| modelId | Specifies the model ID variable name. |
| modelOut | Specifies the output data table to save estimated multitask regression weights. Refer to common `casouttable (Form 1)` for full details (Appendix A: Common Parameters). |
| output | Specifies the output data table to save scored observations. `casOut` subparameter defines the output table details. Refer to common `outputStatement` for full details (Appendix A: Common Parameters). |
| outputTables | Lists the names of results tables to save as CAS tables. `names` subparameter specifies table names. Refer to common `outputTables` for full details (Appendix A: Common Parameters). |
| regL1 | Specifies the L1 (LASSO) penalization weight. Default: 0.01. |
| regL2 | Specifies the L2 graph penalization weight. Default: 0.01. |
| saveState | Specifies the output data table to save the model's state for future scoring. Refer to common `casouttable (Form 1)` for full details (Appendix A: Common Parameters). |
| seed | Specifies the seed for pseudorandom number generation. Default: 0. |
| table | Specifies the input table settings. The table must include one variable named 'id' as the row ID variable. Refer to common `castable (Form 1)` for full details (Appendix A: Common Parameters). |
| targets | Specifies the target variables (required). |
| tolerance | Specifies the optimization tolerance (absolute L2 difference of solution) as a stopping criterion. Default: 1E-06. |