Business Context
A manufacturing plant automates the analysis of machine failure rates. The engineering team wants to archive the 'Model Definition' and 'Fit Summary' of their failure prediction model into CAS tables for a dashboard. Crucially, they want to suppress the visual output in the report (log) to keep the automated pipeline clean, relying solely on the output CAS tables.
Data Preparation
Simulating machine sensor data (temperature, vibration) and failure counts, introducing a few extreme values.
Copied!
| 1 | |
| 2 | DATA casuser.machine_failures; |
| 3 | INPUT temp vibration failures; |
| 4 | DATALINES; |
| 5 | 100 5 0 120 7 1 110 5 0 150 9 5 105 6 0 130 8 2 115 5 1 140 8 3; |
| 6 | |
| 7 | RUN; |
| 8 |
Étapes de réalisation
1
Fit a Poisson model on the machine data and save it to 'failure_store'.
Copied!
| 1 | |
| 2 | PROC CAS; |
| 3 | countreg.countregFitModel / TABLE='machine_failures', model={depVars={{name='failures'}}, effects={{vars={'temp', 'vibration'}}}}, dist='POISSON', store={name='failure_store', replace=true}; |
| 4 | |
| 5 | RUN; |
| 6 |
2
Run ViewStore to output specific tables (ModelInfo, FitSummary) to CAS, while attempting to minimize/control display. Note: We use viewOptions to select specific content and outputTables to save it.
Copied!
| 1 | |
| 2 | PROC CAS; |
| 3 | countreg.countregViewStore / TABLE='machine_failures', instore='failure_store', viewOptions={modelDefinition=true, fitModelSummary=true}, outputTables={names={ModelInfo='archived_info', FitSummary='archived_stats', replace=true}}; |
| 4 | |
| 5 | RUN; |
| 6 |
Expected Result
Two new CAS tables ('archived_info' and 'archived_stats') must be created in the casuser library containing the respective model information. The action should succeed even with the minimal data set provided in the context.