In this exercise, you extract information from the created Classification test result.
db2 -stf heartClasTestResult.db2
SELECT
CAST (IDMMX.DM_getClasError("RESULT") AS DEC(5,2))
AS "CLASSIFICATION_ERROR",'%' AS "%",
CAST (IDMMX.DM_getMdlQuality("RESULT") AS DEC(5,2))
AS "MODEL_QUALITY",
CAST (IDMMX.DM_getReliability("RESULT") AS DEC(5,2))
AS "RELIABILITY",
CAST (IDMMX.DM_getRankQuality("RESULT",'2') AS DEC(5,2))
AS "RANK_QUALITY",
CAST (IDMMX.DM_getPredAccuracy("RESULT",'2') AS DEC(5,2))
AS "PRED_ACCURACY"
FROM IDMMX."CLASTESTRESULTS"
WHERE "ID"='HeartClasTestResult';
The next part of the sample script contains a SELECT statement
that reads the confusion matrix from the result of a Classification
test run. Classifications are divided into actual and predicted values. SELECT CAST ("ACTUAL" AS CHAR(15)) AS "ACTUAL_VALUE",
CAST ("PREDICTED" AS CHAR(15)) AS "PREDICTED_VALUE",
"COUNT" AS "NUM_CASES",
"COST" AS "COST_OF_WRONG_PREDICTION"
FROM TABLE (IDMMX.DM_getConfMatrix(
(SELECT "RESULT" FROM IDMMX."CLASTESTRESULTS"
WHERE ID='HeartClasTestResult'))
) AS "MATRIX"
ORDER BY "ACTUAL", "PREDICTED";
The following part of the sample script contains a SELECT statement
that reads the gains chart information from the result of a Classification
test run: SELECT CAST ("ROWCOUNT" AS SMALLINT) AS "NUMROWS_IN_SUBSET",
CAST ("SUMACTUAL" AS SMALLINT) AS "NUMROWS_DISEASED_IS_2",
CAST ("THRESHOLD" AS DEC(5,2))
AS "SMALLEST_CONFIDENCE_IN_SUBSET"
FROM TABLE (IDMMX.DM_getGainsChart(
(SELECT "RESULT" FROM IDMMX."CLASTESTRESULTS"
WHERE ID='HeartClasTestResult'),'2')
) AS "MATRIX"
ORDER BY "NUMROWS_IN_SUBSET";