Business Context
A large pharmaceutical company monitors the purity level of a drug produced across 100 parallel production lines. To manage this scale, they only want to generate charts for lines that exhibit out-of-control behavior. The analysis must be robust enough to handle a high number of distinct processes.
Data Preparation
Create a large dataset named 'DrugPurity' with 100 production lines ('LineID'), each having 50 batches. Intentionally introduce an out-of-control point in Line 'L-042' and Line 'L-088'.
Copied!
| 1 | DATA mycas.DrugPurity; |
| 2 | LENGTH LineID $ 8; |
| 3 | DO line = 1 to 100; |
| 4 | LineID = 'L-' || put(line, z3.); |
| 5 | DO Batch = 1 to 50; |
| 6 | DO i = 1 to 5; |
| 7 | Purity = 99.5 + 0.1 * rannor(123); |
| 8 | /* Introduce specific outliers */ |
| 9 | IF LineID = 'L-042' and Batch = 30 THEN Purity = 98.9; |
| 10 | IF LineID = 'L-088' and Batch = 15 THEN Purity = 100.2; |
| 11 | OUTPUT; |
| 12 | END; |
| 13 | END; |
| 14 | END; |
| 15 | RUN; |
Étapes de réalisation
1
Run the boxChart action on the large dataset, grouping by 'LineID'. Use 'exChart=True' to only generate chart data for processes with exceptions. Set a high 'groupByLimit' to ensure all 100 lines are processed.
Copied!
| 1 | PROC CAS; |
| 2 | spc.boxChart / |
| 3 | TABLE={name='DrugPurity', groupBy={'LineID'}}, |
| 4 | processValue='Purity', |
| 5 | subgroupValue='Batch', |
| 6 | primaryTests={test1=true}, |
| 7 | exChart=true, |
| 8 | groupByLimit=200, |
| 9 | chartsTable={name='PurityExceptions', replace=true}; |
| 10 | RUN; |
2
Fetch the results from the 'PurityExceptions' table. The results should only contain data for the lines that had an out-of-control point.
Copied!
| 1 | |
| 2 | PROC CAS; |
| 3 | TABLE.fetch / TABLE={name='PurityExceptions'}; |
| 4 | RUN; |
| 5 |
3
Verify that only the expected lines are present in the output table by counting the distinct 'LineID' values.
Copied!
| 1 | PROC CAS; |
| 2 | SIMPLE.distinct / |
| 3 | TABLE={name='PurityExceptions'} |
| 4 | inputs={{name='LineID'}}; |
| 5 | RUN; |
Expected Result
The action should process all 100 lines without hitting a group limit. The resulting 'PurityExceptions' table should be created but should only contain summary data for 'L-042' and 'L-088', as these were the only lines with points outside the 3-sigma limits (failing Test 1). The distinct count on the result table should return 2.