Business Context
A financial institution analyzes transaction sequences to detect fraud patterns using Recurrent Neural Networks (RNN). The automated training pipeline runs nightly. This test ensures that the 'buildModel' action correctly handles the 'replace' parameter when a model table from a previous run already exists, preventing pipeline failures.
About the Set : deepLearn
Creation and training of deep neural networks.
Discover all actions of deepLearn Data Preparation
Simulation of transaction sequence data.
Copied!
| 1 | DATA casuser.transactions; |
| 2 | INPUT account_id time_step amount location_flag; |
| 3 | DATALINES; |
| 4 | 999 1 100.00 0 |
| 5 | 999 2 5000.00 1 |
| 6 | 999 3 20.00 0 |
| 7 | ; |
| 8 | RUN; |
Étapes de réalisation
1
First run: Initial creation of the RNN model.
Copied!
| 1 | PROC CAS; |
| 2 | DEEPLEARN.buildModel / |
| 3 | modelTable={name='fraud_rnn', replace=true} |
| 4 | type='RNN'; |
| 5 | RUN; |
2
Second run: Attempting to overwrite the existing 'fraud_rnn' model to simulate a pipeline rerun.
Copied!
| 1 | PROC CAS; |
| 2 | DEEPLEARN.buildModel / |
| 3 | modelTable={name='fraud_rnn', replace=true} |
| 4 | type='RNN'; |
| 5 | RUN; |
Expected Result
The first step creates the model. The second step successfully overwrites the existing 'fraud_rnn' table without generating an error, confirming that the 'replace=true' parameter functions correctly for automated workflows.