[ML] Data Frame Analytics: ROC Curve Chart (#89991)

Adds the ROC curve chart to the results page for classification jobs in the evaluate section.

x-pack/plugins/ml/public/application/components/scatterplot_matrix/index.ts

@@ -6,6 +6,4 @@
  */
 
 export { useScatterplotFieldOptions } from './use_scatterplot_field_options';
-export { LEGEND_TYPES } from './scatterplot_matrix_vega_lite_spec';
-export { ScatterplotMatrix } from './scatterplot_matrix';
-export type { ScatterplotMatrixViewProps as ScatterplotMatrixProps } from './scatterplot_matrix_view';
+export { ScatterplotMatrix, ScatterplotMatrixProps } from './scatterplot_matrix';

x-pack/plugins/ml/public/application/components/scatterplot_matrix/scatterplot_matrix.tsx

@@ -5,15 +5,305 @@
  * 2.0.
  */
 
-import React, { FC, Suspense } from 'react';
+import React, { useMemo, useEffect, useState, FC } from 'react';
 
-import type { ScatterplotMatrixViewProps } from './scatterplot_matrix_view';
-import { ScatterplotMatrixLoading } from './scatterplot_matrix_loading';
+import {
+  EuiComboBox,
+  EuiComboBoxOptionOption,
+  EuiFlexGroup,
+  EuiFlexItem,
+  EuiFormRow,
+  EuiSelect,
+  EuiSwitch,
+} from '@elastic/eui';
 
-const ScatterplotMatrixLazy = React.lazy(() => import('./scatterplot_matrix_view'));
+import { i18n } from '@kbn/i18n';
 
-export const ScatterplotMatrix: FC<ScatterplotMatrixViewProps> = (props) => (
-  <Suspense fallback={<ScatterplotMatrixLoading />}>
-    <ScatterplotMatrixLazy {...props} />
-  </Suspense>
-);
+import type { SearchResponse7 } from '../../../../common/types/es_client';
+import type { ResultsSearchQuery } from '../../data_frame_analytics/common/analytics';
+
+import { useMlApiContext } from '../../contexts/kibana';
+
+import { getProcessedFields } from '../data_grid';
+import { useCurrentEuiTheme } from '../color_range_legend';
+
+// Separate imports for lazy loadable VegaChart and related code
+import { VegaChart } from '../vega_chart';
+import type { LegendType } from '../vega_chart/common';
+import { VegaChartLoading } from '../vega_chart/vega_chart_loading';
+
+import {
+  getScatterplotMatrixVegaLiteSpec,
+  OUTLIER_SCORE_FIELD,
+} from './scatterplot_matrix_vega_lite_spec';
+
+import './scatterplot_matrix.scss';
+
+const SCATTERPLOT_MATRIX_DEFAULT_FIELDS = 4;
+const SCATTERPLOT_MATRIX_DEFAULT_FETCH_SIZE = 1000;
+const SCATTERPLOT_MATRIX_DEFAULT_FETCH_MIN_SIZE = 1;
+const SCATTERPLOT_MATRIX_DEFAULT_FETCH_MAX_SIZE = 10000;
+
+const TOGGLE_ON = i18n.translate('xpack.ml.splom.toggleOn', {
+  defaultMessage: 'On',
+});
+const TOGGLE_OFF = i18n.translate('xpack.ml.splom.toggleOff', {
+  defaultMessage: 'Off',
+});
+
+const sampleSizeOptions = [100, 1000, 10000].map((d) => ({ value: d, text: '' + d }));
+
+export interface ScatterplotMatrixProps {
+  fields: string[];
+  index: string;
+  resultsField?: string;
+  color?: string;
+  legendType?: LegendType;
+  searchQuery?: ResultsSearchQuery;
+}
+
+export const ScatterplotMatrix: FC<ScatterplotMatrixProps> = ({
+  fields: allFields,
+  index,
+  resultsField,
+  color,
+  legendType,
+  searchQuery,
+}) => {
+  const { esSearch } = useMlApiContext();
+
+  // dynamicSize is optionally used for outlier charts where the scatterplot marks
+  // are sized according to outlier_score
+  const [dynamicSize, setDynamicSize] = useState<boolean>(false);
+
+  // used to give the use the option to customize the fields used for the matrix axes
+  const [fields, setFields] = useState<string[]>([]);
+
+  useEffect(() => {
+    const defaultFields =
+      allFields.length > SCATTERPLOT_MATRIX_DEFAULT_FIELDS
+        ? allFields.slice(0, SCATTERPLOT_MATRIX_DEFAULT_FIELDS)
+        : allFields;
+    setFields(defaultFields);
+  }, [allFields]);
+
+  // the amount of documents to be fetched
+  const [fetchSize, setFetchSize] = useState<number>(SCATTERPLOT_MATRIX_DEFAULT_FETCH_SIZE);
+  // flag to add a random score to the ES query to fetch documents
+  const [randomizeQuery, setRandomizeQuery] = useState<boolean>(false);
+
+  const [isLoading, setIsLoading] = useState<boolean>(false);
+
+  // contains the fetched documents and columns to be passed on to the Vega spec.
+  const [splom, setSplom] = useState<{ items: any[]; columns: string[] } | undefined>();
+
+  // formats the array of field names for EuiComboBox
+  const fieldOptions = useMemo(
+    () =>
+      allFields.map((d) => ({
+        label: d,
+      })),
+    [allFields]
+  );
+
+  const fieldsOnChange = (newFields: EuiComboBoxOptionOption[]) => {
+    setFields(newFields.map((d) => d.label));
+  };
+
+  const fetchSizeOnChange = (e: React.ChangeEvent<HTMLSelectElement>) => {
+    setFetchSize(
+      Math.min(
+        Math.max(parseInt(e.target.value, 10), SCATTERPLOT_MATRIX_DEFAULT_FETCH_MIN_SIZE),
+        SCATTERPLOT_MATRIX_DEFAULT_FETCH_MAX_SIZE
+      )
+    );
+  };
+
+  const randomizeQueryOnChange = () => {
+    setRandomizeQuery(!randomizeQuery);
+  };
+
+  const dynamicSizeOnChange = () => {
+    setDynamicSize(!dynamicSize);
+  };
+
+  const { euiTheme } = useCurrentEuiTheme();
+
+  useEffect(() => {
+    if (fields.length === 0) {
+      setSplom(undefined);
+      setIsLoading(false);
+      return;
+    }
+
+    async function fetchSplom(options: { didCancel: boolean }) {
+      setIsLoading(true);
+      try {
+        const queryFields = [
+          ...fields,
+          ...(color !== undefined ? [color] : []),
+          ...(legendType !== undefined ? [] : [`${resultsField}.${OUTLIER_SCORE_FIELD}`]),
+        ];
+
+        const queryFallback = searchQuery !== undefined ? searchQuery : { match_all: {} };
+        const query = randomizeQuery
+          ? {
+              function_score: {
+                query: queryFallback,
+                random_score: { seed: 10, field: '_seq_no' },
+              },
+            }
+          : queryFallback;
+
+        const resp: SearchResponse7 = await esSearch({
+          index,
+          body: {
+            fields: queryFields,
+            _source: false,
+            query,
+            from: 0,
+            size: fetchSize,
+          },
+        });
+
+        if (!options.didCancel) {
+          const items = resp.hits.hits.map((d) =>
+            getProcessedFields(d.fields, (key: string) =>
+              key.startsWith(`${resultsField}.feature_importance`)
+            )
+          );
+
+          setSplom({ columns: fields, items });
+          setIsLoading(false);
+        }
+      } catch (e) {
+        // TODO error handling
+        setIsLoading(false);
+      }
+    }
+
+    const options = { didCancel: false };
+    fetchSplom(options);
+    return () => {
+      options.didCancel = true;
+    };
+    // stringify the fields array and search, otherwise the comparator will trigger on new but identical instances.
+  }, [fetchSize, JSON.stringify({ fields, searchQuery }), index, randomizeQuery, resultsField]);
+
+  const vegaSpec = useMemo(() => {
+    if (splom === undefined) {
+      return;
+    }
+
+    const { items, columns } = splom;
+
+    const values =
+      resultsField !== undefined
+        ? items
+        : items.map((d) => {
+            d[`${resultsField}.${OUTLIER_SCORE_FIELD}`] = 0;
+            return d;
+          });
+
+    return getScatterplotMatrixVegaLiteSpec(
+      values,
+      columns,
+      euiTheme,
+      resultsField,
+      color,
+      legendType,
+      dynamicSize
+    );
+  }, [resultsField, splom, color, legendType, dynamicSize]);
+
+  return (
+    <>
+      {splom === undefined || vegaSpec === undefined ? (
+        <VegaChartLoading />
+      ) : (
+        <div data-test-subj="mlScatterplotMatrix">
+          <EuiFlexGroup>
+            <EuiFlexItem>
+              <EuiFormRow
+                label={i18n.translate('xpack.ml.splom.fieldSelectionLabel', {
+                  defaultMessage: 'Fields',
+                })}
+                display="rowCompressed"
+                fullWidth
+              >
+                <EuiComboBox
+                  compressed
+                  fullWidth
+                  placeholder={i18n.translate('xpack.ml.splom.fieldSelectionPlaceholder', {
+                    defaultMessage: 'Select fields',
+                  })}
+                  options={fieldOptions}
+                  selectedOptions={fields.map((d) => ({
+                    label: d,
+                  }))}
+                  onChange={fieldsOnChange}
+                  isClearable={true}
+                  data-test-subj="mlScatterplotMatrixFieldsComboBox"
+                />
+              </EuiFormRow>
+            </EuiFlexItem>
+            <EuiFlexItem style={{ width: '200px' }} grow={false}>
+              <EuiFormRow
+                label={i18n.translate('xpack.ml.splom.sampleSizeLabel', {
+                  defaultMessage: 'Sample size',
+                })}
+                display="rowCompressed"
+                fullWidth
+              >
+                <EuiSelect
+                  compressed
+                  options={sampleSizeOptions}
+                  value={fetchSize}
+                  onChange={fetchSizeOnChange}
+                />
+              </EuiFormRow>
+            </EuiFlexItem>
+            <EuiFlexItem style={{ width: '120px' }} grow={false}>
+              <EuiFormRow
+                label={i18n.translate('xpack.ml.splom.randomScoringLabel', {
+                  defaultMessage: 'Random scoring',
+                })}
+                display="rowCompressed"
+                fullWidth
+              >
+                <EuiSwitch
+                  name="mlScatterplotMatrixRandomizeQuery"
+                  label={randomizeQuery ? TOGGLE_ON : TOGGLE_OFF}
+                  checked={randomizeQuery}
+                  onChange={randomizeQueryOnChange}
+                  disabled={isLoading}
+                />
+              </EuiFormRow>
+            </EuiFlexItem>
+            {resultsField !== undefined && legendType === undefined && (
+              <EuiFlexItem style={{ width: '120px' }} grow={false}>
+                <EuiFormRow
+                  label={i18n.translate('xpack.ml.splom.dynamicSizeLabel', {
+                    defaultMessage: 'Dynamic size',
+                  })}
+                  display="rowCompressed"
+                  fullWidth
+                >
+                  <EuiSwitch
+                    name="mlScatterplotMatrixDynamicSize"
+                    label={dynamicSize ? TOGGLE_ON : TOGGLE_OFF}
+                    checked={dynamicSize}
+                    onChange={dynamicSizeOnChange}
+                    disabled={isLoading}
+                  />
+                </EuiFormRow>
+              </EuiFlexItem>
+            )}
+          </EuiFlexGroup>
+
+          <VegaChart vegaSpec={vegaSpec} />
+        </div>
+      )}
+    </>
+  );
+};

x-pack/plugins/ml/public/application/components/scatterplot_matrix/scatterplot_matrix_vega_lite_spec.test.ts

@@ -10,13 +10,14 @@ import { compile } from 'vega-lite/build-es5/vega-lite';
 
 import euiThemeLight from '@elastic/eui/dist/eui_theme_light.json';
 
+import { LEGEND_TYPES } from '../vega_chart/common';
+
 import {
   getColorSpec,
   getScatterplotMatrixVegaLiteSpec,
   COLOR_OUTLIER,
   COLOR_RANGE_NOMINAL,
   DEFAULT_COLOR,
-  LEGEND_TYPES,
 } from './scatterplot_matrix_vega_lite_spec';
 
 describe('getColorSpec()', () => {

x-pack/plugins/ml/public/application/components/scatterplot_matrix/scatterplot_matrix_vega_lite_spec.ts

@@ -15,11 +15,7 @@ import { euiPaletteColorBlind, euiPaletteNegative, euiPalettePositive } from '@e
 
 import { i18n } from '@kbn/i18n';
 
-export const LEGEND_TYPES = {
-  NOMINAL: 'nominal',
-  QUANTITATIVE: 'quantitative',
-} as const;
-export type LegendType = typeof LEGEND_TYPES[keyof typeof LEGEND_TYPES];
+import { LegendType, LEGEND_TYPES } from '../vega_chart/common';
 
 export const OUTLIER_SCORE_FIELD = 'outlier_score';