Contexto empresarial
A financial institution needs to classify its customers into risk categories based on their credit score. This allows for tailored product offerings and risk management. The format should be simple, robust, and easily updatable.
Preparación de datos
Creation of a small customer dataset with credit scores. This table will be loaded into CAS to test the format application.
¡Copiado!
| 1 | DATA casuser.CustomerScores; |
| 2 | INFILE DATALINES; |
| 3 | INPUT CustomerID Score; |
| 4 | DATALINES; |
| 5 | 101 820 |
| 6 | 102 745 |
| 7 | 103 650 |
| 8 | 104 580 |
| 9 | 105 999 |
| 10 | 106 350 |
| 11 | ; |
| 12 | RUN; |
Étapes de réalisation
1
Create a new format library named 'RiskFormats' and ensure it's clean by using the 'replace' option.
¡Copiado!
| 1 | PROC CAS; |
| 2 | ACTION sessionProp.addFmtLib / |
| 3 | fmtLibName='RiskFormats', |
| 4 | replace=true; |
| 5 | RUN; |
| 6 | QUIT; |
2
Define a numeric format 'ScoreFmt' to map credit scores to risk labels. The 'other' keyword handles unexpected values.
¡Copiado!
| 1 | PROC CAS; |
| 2 | ACTION sessionProp.addFormat / |
| 3 | fmtLibName='RiskFormats' |
| 4 | fmtName='ScoreFmt' |
| 5 | ranges={'300-579=High Risk', '580-669=Fair Risk', '670-799=Good Risk', '800-850=Excellent', 'other=Review'}; |
| 6 | RUN; |
| 7 | QUIT; |
3
Apply the newly created format to the 'Score' column and display the results to verify the mapping.
¡Copiado!
| 1 | PROC casutil; |
| 2 | load DATA=casuser.CustomerScores outcaslib='casuser' casout='CustomerScores' replace; |
| 3 | QUIT; |
| 4 | PROC CAS; |
| 5 | TABLE.fetch / |
| 6 | TABLE='CustomerScores', |
| 7 | FORMAT=true, |
| 8 | vars={'CustomerID', 'Score'}, |
| 9 | formatDef={name='Score', FORMAT='ScoreFmt'}; |
| 10 | RUN; |
| 11 | QUIT; |
Resultado esperado
The final fetched table should display a new formatted column for 'Score'. For example, customer 101 with a score of 820 should display 'Excellent', customer 103 with 650 should display 'Fair Risk', and customer 105 with 999 should display 'Review', demonstrating the 'other' keyword works correctly.