[ML] Fixing categorization wizard example results (#54924) (#55438)

* [ML] Fixing categorization wizard example results * moving validation results to class * cleaning up category analyzer types * small tweaks * removing commented out code * fixing string ids * small refactor * improving validation messages * fixing types * updating message text * fixing typo * adding privileges error * updating privilege message * changes based on review * removing old warning message * fixing translations * renaming enum
elastic · Jan 21, 2020 · 15e584f · 15e584f
1 parent c96ca20
commit 15e584f
Show file tree

Hide file tree

Showing 20 changed files with 763 additions and 420 deletions.
diff --git a/x-pack/legacy/plugins/ml/common/constants/new_job.ts b/x-pack/legacy/plugins/ml/common/constants/new_job.ts
@@ -26,7 +26,14 @@ export const DEFAULT_QUERY_DELAY = '60s';
 
 export const SHARED_RESULTS_INDEX_NAME = 'shared';
 
+// Categorization
 export const NUMBER_OF_CATEGORY_EXAMPLES = 5;
 export const CATEGORY_EXAMPLES_SAMPLE_SIZE = 1000;
 export const CATEGORY_EXAMPLES_WARNING_LIMIT = 0.75;
 export const CATEGORY_EXAMPLES_ERROR_LIMIT = 0.02;
+
+export enum CATEGORY_EXAMPLES_VALIDATION_STATUS {
+  VALID = 'valid',
+  PARTIALLY_VALID = 'partially_valid',
+  INVALID = 'invalid',
+}
diff --git a/x-pack/legacy/plugins/ml/common/types/categories.ts b/x-pack/legacy/plugins/ml/common/types/categories.ts
@@ -4,6 +4,8 @@
  * you may not use this file except in compliance with the Elastic License.
  */
 
+import { CATEGORY_EXAMPLES_VALIDATION_STATUS } from '../constants/new_job';
+
 export type CategoryId = number;
 
 export interface Category {
@@ -23,3 +25,30 @@ export interface Token {
   type: string;
   position: number;
 }
+
+export interface CategorizationAnalyzer {
+  char_filter?: any[];
+  tokenizer?: string;
+  filter?: any[];
+  analyzer?: string;
+}
+
+export interface CategoryFieldExample {
+  text: string;
+  tokens: Token[];
+}
+
+export enum VALIDATION_RESULT {
+  TOKEN_COUNT,
+  MEDIAN_LINE_LENGTH,
+  NULL_VALUES,
+  TOO_MANY_TOKENS,
+  FAILED_TO_TOKENIZE,
+  INSUFFICIENT_PRIVILEGES,
+}
+
+export interface FieldExampleCheck {
+  id: VALIDATION_RESULT;
+  valid: CATEGORY_EXAMPLES_VALIDATION_STATUS;
+  message: string;
+}
diff --git a/x-pack/legacy/plugins/ml/common/util/string_utils.test.ts b/x-pack/legacy/plugins/ml/common/util/string_utils.test.ts
@@ -4,7 +4,17 @@
  * you may not use this file except in compliance with the Elastic License.
  */
 
-import { renderTemplate } from './string_utils';
+import { renderTemplate, getMedianStringLength } from './string_utils';
+
+const strings: string[] = [
+  'foo',
+  'foofoofoofoofoo',
+  'foofoofoo',
+  'f',
+  'f',
+  'foofoofoofoofoofoofoo',
+];
+const noStrings: string[] = [];
 
 describe('ML - string utils', () => {
   describe('renderTemplate', () => {
@@ -24,4 +34,16 @@ describe('ML - string utils', () => {
       expect(result).toBe('string with 1 replacement, and a 2nd one.');
     });
   });
+
+  describe('getMedianStringLength', () => {
+    test('test median for string array', () => {
+      const result = getMedianStringLength(strings);
+      expect(result).toBe(9);
+    });
+
+    test('test median for no strings', () => {
+      const result = getMedianStringLength(noStrings);
+      expect(result).toBe(0);
+    });
+  });
 });
diff --git a/x-pack/legacy/plugins/ml/common/util/string_utils.ts b/x-pack/legacy/plugins/ml/common/util/string_utils.ts
@@ -17,3 +17,8 @@ export function renderTemplate(str: string, data?: Record<string, string>): stri
 
   return str;
 }
+
+export function getMedianStringLength(strings: string[]) {
+  const sortedStringLengths = strings.map(s => s.length).sort((a, b) => a - b);
+  return sortedStringLengths[Math.floor(sortedStringLengths.length / 2)] || 0;
+}
diff --git a/...ugins/ml/public/application/jobs/new_job/common/job_creator/categorization_job_creator.ts b/...ugins/ml/public/application/jobs/new_job/common/job_creator/categorization_job_creator.ts
@@ -16,25 +16,31 @@ import {
   CREATED_BY_LABEL,
   DEFAULT_BUCKET_SPAN,
   DEFAULT_RARE_BUCKET_SPAN,
+  CATEGORY_EXAMPLES_VALIDATION_STATUS,
 } from '../../../../../../common/constants/new_job';
 import { ML_JOB_AGGREGATION } from '../../../../../../common/constants/aggregation_types';
+import {
+  CategorizationAnalyzer,
+  CategoryFieldExample,
+  FieldExampleCheck,
+} from '../../../../../../common/types/categories';
 import { getRichDetectors } from './util/general';
-import { CategorizationExamplesLoader, CategoryExample } from '../results_loader';
-import { CategorizationAnalyzer, getNewJobDefaults } from '../../../../services/ml_server_info';
-
-type CategorizationAnalyzerType = CategorizationAnalyzer | null;
+import { CategorizationExamplesLoader } from '../results_loader';
+import { getNewJobDefaults } from '../../../../services/ml_server_info';
 
 export class CategorizationJobCreator extends JobCreator {
   protected _type: JOB_TYPE = JOB_TYPE.CATEGORIZATION;
   private _createCountDetector: () => void = () => {};
   private _createRareDetector: () => void = () => {};
   private _examplesLoader: CategorizationExamplesLoader;
-  private _categoryFieldExamples: CategoryExample[] = [];
-  private _categoryFieldValid: number = 0;
+  private _categoryFieldExamples: CategoryFieldExample[] = [];
+  private _validationChecks: FieldExampleCheck[] = [];
+  private _overallValidStatus: CATEGORY_EXAMPLES_VALIDATION_STATUS =
+    CATEGORY_EXAMPLES_VALIDATION_STATUS.INVALID;
   private _detectorType: ML_JOB_AGGREGATION.COUNT | ML_JOB_AGGREGATION.RARE =
     ML_JOB_AGGREGATION.COUNT;
-  private _categorizationAnalyzer: CategorizationAnalyzerType = null;
-  private _defaultCategorizationAnalyzer: CategorizationAnalyzerType;
+  private _categorizationAnalyzer: CategorizationAnalyzer = {};
+  private _defaultCategorizationAnalyzer: CategorizationAnalyzer;
 
   constructor(
     indexPattern: IndexPattern,
@@ -46,7 +52,7 @@ export class CategorizationJobCreator extends JobCreator {
     this._examplesLoader = new CategorizationExamplesLoader(this, indexPattern, query);
 
     const { anomaly_detectors: anomalyDetectors } = getNewJobDefaults();
-    this._defaultCategorizationAnalyzer = anomalyDetectors.categorization_analyzer || null;
+    this._defaultCategorizationAnalyzer = anomalyDetectors.categorization_analyzer || {};
   }
 
   public setDefaultDetectorProperties(
@@ -93,7 +99,7 @@ export class CategorizationJobCreator extends JobCreator {
     } else {
       delete this._job_config.analysis_config.categorization_field_name;
       this._categoryFieldExamples = [];
-      this._categoryFieldValid = 0;
+      this._validationChecks = [];
     }
   }
 
@@ -102,31 +108,38 @@ export class CategorizationJobCreator extends JobCreator {
   }
 
   public async loadCategorizationFieldExamples() {
-    const { valid, examples, sampleSize } = await this._examplesLoader.loadExamples();
+    const {
+      examples,
+      sampleSize,
+      overallValidStatus,
+      validationChecks,
+    } = await this._examplesLoader.loadExamples();
     this._categoryFieldExamples = examples;
-    this._categoryFieldValid = valid;
-    return { valid, examples, sampleSize };
+    this._validationChecks = validationChecks;
+    this._overallValidStatus = overallValidStatus;
+    return { examples, sampleSize, overallValidStatus, validationChecks };
   }
 
   public get categoryFieldExamples() {
     return this._categoryFieldExamples;
   }
 
-  public get categoryFieldValid() {
-    return this._categoryFieldValid;
+  public get validationChecks() {
+    return this._validationChecks;
+  }
+
+  public get overallValidStatus() {
+    return this._overallValidStatus;
   }
 
   public get selectedDetectorType() {
     return this._detectorType;
   }
 
-  public set categorizationAnalyzer(analyzer: CategorizationAnalyzerType) {
+  public set categorizationAnalyzer(analyzer: CategorizationAnalyzer) {
     this._categorizationAnalyzer = analyzer;
 
-    if (
-      analyzer === null ||
-      isEqual(this._categorizationAnalyzer, this._defaultCategorizationAnalyzer)
-    ) {
+    if (isEqual(this._categorizationAnalyzer, this._defaultCategorizationAnalyzer)) {
       delete this._job_config.analysis_config.categorization_analyzer;
     } else {
       this._job_config.analysis_config.categorization_analyzer = analyzer;

diff --git a/...k/legacy/plugins/ml/public/application/jobs/new_job/common/job_validator/job_validator.ts b/...k/legacy/plugins/ml/public/application/jobs/new_job/common/job_validator/job_validator.ts
@@ -16,7 +16,7 @@ import { JobCreator, JobCreatorType, isCategorizationJobCreator } from '../job_c
 import { populateValidationMessages, checkForExistingJobAndGroupIds } from './util';
 import { ExistingJobsAndGroups } from '../../../../services/job_service';
 import { cardinalityValidator, CardinalityValidatorResult } from './validators';
-import { CATEGORY_EXAMPLES_ERROR_LIMIT } from '../../../../../../common/constants/new_job';
+import { CATEGORY_EXAMPLES_VALIDATION_STATUS } from '../../../../../../common/constants/new_job';
 
 // delay start of validation to allow the user to make changes
 // e.g. if they are typing in a new value, try not to validate
@@ -207,7 +207,7 @@ export class JobValidator {
   private _runAdvancedValidation() {
     if (isCategorizationJobCreator(this._jobCreator)) {
       this._advancedValidations.categorizationFieldValid.valid =
-        this._jobCreator.categoryFieldValid > CATEGORY_EXAMPLES_ERROR_LIMIT;
+        this._jobCreator.overallValidStatus !== CATEGORY_EXAMPLES_VALIDATION_STATUS.INVALID;
     }
   }
 

diff --git a/...l/public/application/jobs/new_job/common/results_loader/categorization_examples_loader.ts b/...l/public/application/jobs/new_job/common/results_loader/categorization_examples_loader.ts
@@ -6,15 +6,12 @@
 
 import { IndexPattern } from '../../../../../../../../../../src/plugins/data/public';
 import { IndexPatternTitle } from '../../../../../../common/types/kibana';
-import { Token } from '../../../../../../common/types/categories';
 import { CategorizationJobCreator } from '../job_creator';
 import { ml } from '../../../../services/ml_api_service';
-import { NUMBER_OF_CATEGORY_EXAMPLES } from '../../../../../../common/constants/new_job';
-
-export interface CategoryExample {
-  text: string;
-  tokens: Token[];
-}
+import {
+  NUMBER_OF_CATEGORY_EXAMPLES,
+  CATEGORY_EXAMPLES_VALIDATION_STATUS,
+} from '../../../../../../common/constants/new_job';
 
 export class CategorizationExamplesLoader {
   private _jobCreator: CategorizationJobCreator;
@@ -36,20 +33,22 @@ export class CategorizationExamplesLoader {
     const analyzer = this._jobCreator.categorizationAnalyzer;
     const categorizationFieldName = this._jobCreator.categorizationFieldName;
     if (categorizationFieldName === null) {
-      return { valid: 0, examples: [], sampleSize: 0 };
+      return {
+        examples: [],
+        sampleSize: 0,
+        overallValidStatus: CATEGORY_EXAMPLES_VALIDATION_STATUS.INVALID,
+        validationChecks: [],
+      };
     }
 
-    const start = Math.floor(
-      this._jobCreator.start + (this._jobCreator.end - this._jobCreator.start) / 2
-    );
     const resp = await ml.jobs.categorizationFieldExamples(
       this._indexPatternTitle,
       this._query,
       NUMBER_OF_CATEGORY_EXAMPLES,
       categorizationFieldName,
       this._timeFieldName,
-      start,
-      0,
+      this._jobCreator.start,
+      this._jobCreator.end,
       analyzer
     );
     return resp;

diff --git a/x-pack/legacy/plugins/ml/public/application/jobs/new_job/common/results_loader/index.ts b/x-pack/legacy/plugins/ml/public/application/jobs/new_job/common/results_loader/index.ts
@@ -5,4 +5,4 @@
  */
 
 export { ResultsLoader, Results, ModelItem, Anomaly } from './results_loader';
-export { CategorizationExamplesLoader, CategoryExample } from './categorization_examples_loader';
+export { CategorizationExamplesLoader } from './categorization_examples_loader';
diff --git a/...ges/components/pick_fields_step/components/categorization_view/examples_valid_callout.tsx b/...ges/components/pick_fields_step/components/categorization_view/examples_valid_callout.tsx
@@ -9,27 +9,24 @@ import { EuiCallOut, EuiSpacer, EuiCallOutProps } from '@elastic/eui';
 import { i18n } from '@kbn/i18n';
 import { FormattedMessage } from '@kbn/i18n/react';
 
-import { CategorizationAnalyzer } from '../../../../../../../services/ml_server_info';
-import { EditCategorizationAnalyzerFlyout } from '../../../common/edit_categorization_analyzer_flyout';
 import {
-  CATEGORY_EXAMPLES_ERROR_LIMIT,
-  CATEGORY_EXAMPLES_WARNING_LIMIT,
-} from '../../../../../../../../../common/constants/new_job';
-
-type CategorizationAnalyzerType = CategorizationAnalyzer | null;
+  CategorizationAnalyzer,
+  FieldExampleCheck,
+} from '../../../../../../../../../common/types/categories';
+import { EditCategorizationAnalyzerFlyout } from '../../../common/edit_categorization_analyzer_flyout';
+import { CATEGORY_EXAMPLES_VALIDATION_STATUS } from '../../../../../../../../../common/constants/new_job';
 
 interface Props {
-  examplesValid: number;
-  sampleSize: number;
-  categorizationAnalyzer: CategorizationAnalyzerType;
+  validationChecks: FieldExampleCheck[];
+  overallValidStatus: CATEGORY_EXAMPLES_VALIDATION_STATUS;
+  categorizationAnalyzer: CategorizationAnalyzer;
 }
 
 export const ExamplesValidCallout: FC<Props> = ({
-  examplesValid,
+  overallValidStatus,
+  validationChecks,
   categorizationAnalyzer,
-  sampleSize,
 }) => {
-  const percentageText = <PercentageText examplesValid={examplesValid} sampleSize={sampleSize} />;
   const analyzerUsed = <AnalyzerUsed categorizationAnalyzer={categorizationAnalyzer} />;
 
   let color: EuiCallOutProps['color'] = 'success';
@@ -40,15 +37,15 @@ export const ExamplesValidCallout: FC<Props> = ({
     }
   );
 
-  if (examplesValid < CATEGORY_EXAMPLES_ERROR_LIMIT) {
+  if (overallValidStatus === CATEGORY_EXAMPLES_VALIDATION_STATUS.INVALID) {
     color = 'danger';
     title = i18n.translate(
       'xpack.ml.newJob.wizard.pickFieldsStep.categorizationFieldCalloutTitle.invalid',
       {
         defaultMessage: 'Selected category field is invalid',
       }
     );
-  } else if (examplesValid < CATEGORY_EXAMPLES_WARNING_LIMIT) {
+  } else if (overallValidStatus === CATEGORY_EXAMPLES_VALIDATION_STATUS.PARTIALLY_VALID) {
     color = 'warning';
     title = i18n.translate(
       'xpack.ml.newJob.wizard.pickFieldsStep.categorizationFieldCalloutTitle.possiblyInvalid',
@@ -60,45 +57,24 @@ export const ExamplesValidCallout: FC<Props> = ({
 
   return (
     <EuiCallOut color={color} title={title}>
-      {percentageText}
+      {validationChecks.map((v, i) => (
+        <div key={i}>{v.message}</div>
+      ))}
       <EuiSpacer size="s" />
       {analyzerUsed}
     </EuiCallOut>
   );
 };
 
-const PercentageText: FC<{ examplesValid: number; sampleSize: number }> = ({
-  examplesValid,
-  sampleSize,
-}) => (
-  <div>
-    <FormattedMessage
-      id="xpack.ml.newJob.wizard.pickFieldsStep.categorizationFieldPercentage"
-      defaultMessage="{number} field {number, plural, zero {value} one {value} other {values}} analyzed, {percentage}% contain valid tokens."
-      values={{
-        number: sampleSize,
-        percentage: Math.floor(examplesValid * 100),
-      }}
-    />
-  </div>
-);
-
-const AnalyzerUsed: FC<{ categorizationAnalyzer: CategorizationAnalyzerType }> = ({
+const AnalyzerUsed: FC<{ categorizationAnalyzer: CategorizationAnalyzer }> = ({
   categorizationAnalyzer,
 }) => {
   let analyzer = '';
-  if (typeof categorizationAnalyzer === null) {
-    return null;
-  }
 
-  if (typeof categorizationAnalyzer === 'string') {
-    analyzer = categorizationAnalyzer;
-  } else {
-    if (categorizationAnalyzer?.tokenizer !== undefined) {
-      analyzer = categorizationAnalyzer?.tokenizer!;
-    } else if (categorizationAnalyzer?.analyzer !== undefined) {
-      analyzer = categorizationAnalyzer?.analyzer!;
-    }
+  if (categorizationAnalyzer?.tokenizer !== undefined) {
+    analyzer = categorizationAnalyzer.tokenizer;
+  } else if (categorizationAnalyzer?.analyzer !== undefined) {
+    analyzer = categorizationAnalyzer.analyzer;
   }
 
   return (