|
@@ -348,16 +348,20 @@ public class ClassificationIT extends MlNativeDataFrameAnalyticsIntegTestCase {
|
|
|
assertEvaluation(KEYWORD_FIELD, KEYWORD_FIELD_VALUES, "ml." + predictedClassField);
|
|
|
}
|
|
|
|
|
|
- @AwaitsFix(bugUrl = "Muted until ml-cpp supports multiple classes")
|
|
|
public void testDependentVariableCardinalityTooHighError() throws Exception {
|
|
|
initialize("cardinality_too_high");
|
|
|
indexData(sourceIndex, 6, 5, KEYWORD_FIELD);
|
|
|
|
|
|
- // Index one more document with a class different than the two already used.
|
|
|
- client().execute(IndexAction.INSTANCE, new IndexRequest(sourceIndex)
|
|
|
- .source(KEYWORD_FIELD, "fox")
|
|
|
- .setRefreshPolicy(WriteRequest.RefreshPolicy.IMMEDIATE))
|
|
|
- .actionGet();
|
|
|
+ // Index enough documents to have more classes than the allowed limit
|
|
|
+ BulkRequestBuilder bulkRequestBuilder = client().prepareBulk().setRefreshPolicy(WriteRequest.RefreshPolicy.IMMEDIATE);
|
|
|
+ for (int i = 0; i < Classification.MAX_DEPENDENT_VARIABLE_CARDINALITY - 1; i++) {
|
|
|
+ IndexRequest indexRequest = new IndexRequest(sourceIndex).source(KEYWORD_FIELD, "fox-" + i);
|
|
|
+ bulkRequestBuilder.add(indexRequest);
|
|
|
+ }
|
|
|
+ BulkResponse bulkResponse = bulkRequestBuilder.get();
|
|
|
+ if (bulkResponse.hasFailures()) {
|
|
|
+ fail("Failed to index data: " + bulkResponse.buildFailureMessage());
|
|
|
+ }
|
|
|
|
|
|
DataFrameAnalyticsConfig config = buildAnalytics(jobId, sourceIndex, destIndex, null, new Classification(KEYWORD_FIELD));
|
|
|
registerAnalytics(config);
|
|
@@ -365,7 +369,7 @@ public class ClassificationIT extends MlNativeDataFrameAnalyticsIntegTestCase {
|
|
|
|
|
|
ElasticsearchStatusException e = expectThrows(ElasticsearchStatusException.class, () -> startAnalytics(jobId));
|
|
|
assertThat(e.status().getStatus(), equalTo(400));
|
|
|
- assertThat(e.getMessage(), equalTo("Field [keyword-field] must have at most [2] distinct values but there were at least [3]"));
|
|
|
+ assertThat(e.getMessage(), equalTo("Field [keyword-field] must have at most [30] distinct values but there were at least [31]"));
|
|
|
}
|
|
|
|
|
|
public void testDependentVariableCardinalityTooHighButWithQueryMakesItWithinRange() throws Exception {
|