Nate/linear regressions (#290)

Scatter chart fixes for regression lines

web/src/components/ui/charts/BusterChartJS/hooks/useSeriesOptions/scatterSeriesBuilder.test.ts

@@ -1,14 +1,11 @@
 import { type IColumnLabelFormat } from '@/api/asset_interfaces/metric';
 import { scatterSeriesBuilder_data, scatterSeriesBuilder_labels } from './scatterSeriesBuilder';
 import { createDayjsDate } from '@/lib/date';
-import type {
+import type { DatasetOptionsWithTicks } from '../../../chartHooks/useDatasetOptions/interfaces';
-  DatasetOptionsWithTicks,
-  DatasetOption
-} from '../../../chartHooks/useDatasetOptions/interfaces';
 import type { SimplifiedColumnType } from '@/api/asset_interfaces/metric';
 import type { SeriesBuilderProps } from './interfaces';
-import { ChartType } from '@/api/asset_interfaces/metric/charts/enum';
 import type { LabelBuilderProps } from './useSeriesOptions';
+import { DEFAULT_COLUMN_LABEL_FORMAT } from '@/api/asset_interfaces/metric';
 
 describe('scatterSeriesBuilder_data', () => {
   const mockColors = ['#FF0000', '#00FF00'];
@@ -126,67 +123,118 @@ describe('scatterSeriesBuilder_data', () => {
 });
 
 describe('scatterSeriesBuilder_labels', () => {
-  const baseTrendlineSeries = [
+  test('should return undefined when trendlineSeries is empty', () => {
-    {
+    const props: LabelBuilderProps = {
-      yAxisKey: 'metric1',
+      trendlineSeries: [],
-      data: [10, 20, 30],
+      datasetOptions: {
-      label: 'Trendline 1',
+        ticks: [],
-      tooltipData: [],
+        datasets: [],
-      xAxisKeys: ['timestamp'],
+        ticksKey: []
-      type: 'line' as const,
+      },
-      borderColor: '#FF0000',
+      columnLabelFormats: {},
-      borderWidth: 2,
+      xAxisKeys: ['x'],
-      pointRadius: 0
+      sizeKey: [],
-    }
+      columnSettings: {}
-  ];
+    };
-
-  const baseProps: LabelBuilderProps = {
-    trendlineSeries: baseTrendlineSeries,
-    datasetOptions: {
-      datasets: [
-        {
-          id: '1',
-          dataKey: 'metric1',
-          data: [10, 20, 30],
-          ticksForScatter: [
-            [1000, 'Jan 1'],
-            [2000, 'Jan 2'],
-            [3000, 'Jan 3']
-          ],
-          label: [{ key: 'metric1', value: 'Dataset 1' }],
-          axisType: 'y',
-          tooltipData: []
-        }
-      ],
-      ticks: [
-        [1000, 'Jan 1'],
-        [2000, 'Jan 2'],
-        [3000, 'Jan 3']
-      ],
-      ticksKey: [{ key: 'timestamp', value: 'Timestamp' }]
-    },
-    columnLabelFormats: {
-      timestamp: {
-        columnType: 'timestamp' as SimplifiedColumnType,
-        style: 'date'
-      }
-    },
-    xAxisKeys: ['timestamp'],
-    sizeKey: [],
-    columnSettings: {}
-  };
-
-  it('should return undefined when no trendlines exist', () => {
-    const result = scatterSeriesBuilder_labels({
-      ...baseProps,
-      trendlineSeries: []
-    });
 
+    const result = scatterSeriesBuilder_labels(props);
     expect(result).toBeUndefined();
   });
 
-  it('should process trendline series correctly', () => {
+  test('should process date labels directly when x-axis is date type', () => {
-    const result = scatterSeriesBuilder_labels(baseProps);
+    const dateString1 = '2023-01-01';
-    expect(result).toBeDefined();
+    const dateString2 = '2023-01-02';
+
+    const props: LabelBuilderProps = {
+      trendlineSeries: [{ yAxisKey: 'y1' } as any],
+      datasetOptions: {
+        ticks: [[dateString1], [dateString2]],
+        datasets: [{ dataKey: 'y1', data: [10, 20] } as any],
+        ticksKey: [{ key: 'x', value: 'X Axis' }]
+      } as any,
+      columnLabelFormats: {
+        x: {
+          columnType: 'date',
+          style: 'date'
+        }
+      },
+      xAxisKeys: ['x'],
+      sizeKey: [],
+      columnSettings: {}
+    };
+
+    const result = scatterSeriesBuilder_labels(props);
+    expect(result).toEqual([
+      createDayjsDate(dateString1).toDate(),
+      createDayjsDate(dateString2).toDate()
+    ]);
+  });
+
+  test('should collect all ticks without deduplication', () => {
+    const props: LabelBuilderProps = {
+      trendlineSeries: [{ yAxisKey: 'y1' } as any, { yAxisKey: 'y2' } as any],
+      datasetOptions: {
+        ticks: [],
+        datasets: [
+          {
+            dataKey: 'y1',
+            data: [10, 20],
+            ticksForScatter: [
+              [1, 'A'],
+              [2, 'B']
+            ]
+          } as any,
+          {
+            dataKey: 'y2',
+            data: [30, 40, 50],
+            ticksForScatter: [
+              [1, 'A'],
+              [3, 'C'],
+              [5, 'D']
+            ]
+          } as any
+        ],
+        ticksKey: [{ key: 'x', value: 'X Axis' }]
+      } as any,
+      columnLabelFormats: {
+        x: DEFAULT_COLUMN_LABEL_FORMAT
+      },
+      xAxisKeys: ['x'],
+      sizeKey: [],
+      columnSettings: {}
+    };
+
+    const result = scatterSeriesBuilder_labels(props);
+    // Should include duplicate [1, 'A'] from both datasets
+    expect(result).toEqual([1, 'A', 1, 'A', 2, 'B', 3, 'C', 5, 'D']);
+  });
+
+  test('should return undefined when no relevant datasets are found', () => {
+    const props: LabelBuilderProps = {
+      trendlineSeries: [{ yAxisKey: 'y1' } as any],
+      datasetOptions: {
+        ticks: [],
+        datasets: [
+          {
+            dataKey: 'y2', // Not matching with trendlineSeries
+            data: [10, 20],
+            ticksForScatter: [
+              [1, 'A'],
+              [2, 'B']
+            ]
+          } as any
+        ],
+        ticksKey: [{ key: 'x', value: 'X Axis' }]
+      } as any,
+      columnLabelFormats: {
+        x: DEFAULT_COLUMN_LABEL_FORMAT
+      },
+      xAxisKeys: ['x'],
+      sizeKey: [],
+      columnSettings: {}
+    };
+
+    const result = scatterSeriesBuilder_labels(props);
+    expect(result).toBeUndefined();
   });
 });

web/src/components/ui/charts/BusterChartJS/hooks/useSeriesOptions/scatterSeriesBuilder.ts

@@ -143,33 +143,44 @@ const computeSizeRatio = (
 };
 
 export const scatterSeriesBuilder_labels = (props: LabelBuilderProps) => {
-  const { trendlineSeries, datasetOptions } = props;
+  const { trendlineSeries, datasetOptions, columnLabelFormats, xAxisKeys } = props;
 
-  if (trendlineSeries.length > 0) {
+  if (!trendlineSeries.length) return undefined;
-    // Create a Set of relevant yAxisKeys for O(1) lookup
-    const relevantYAxisKeys = new Set(trendlineSeries.map((t) => t.yAxisKey));
 
-    // Combine filtering, flattening and uniqueness in a single pass
+  // Create a Set of relevant yAxisKeys for O(1) lookup
-    const allTicksForScatter = datasetOptions.datasets
+  const relevantYAxisKeys = new Set(trendlineSeries.map((t) => t.yAxisKey));
-      .filter((dataset) => relevantYAxisKeys.has(dataset.dataKey))
-      .flatMap((dataset) => dataset.ticksForScatter || [])
-      .sort((a, b) => {
-        const aVal = a[0],
-          bVal = b[0];
-        return aVal < bVal ? -1 : aVal > bVal ? 1 : 0;
-      })
-      .filter((tick, index, array) => index === 0 || tick[0] !== array[index - 1][0]);
 
-    const modifyFiedProps = {
+  // Get X-axis format information once
-      ...props,
+  const xColumnLabelFormat = columnLabelFormats[xAxisKeys[0]] || DEFAULT_COLUMN_LABEL_FORMAT;
-      datasetOptions: {
+  const useDateLabels =
-        ...props.datasetOptions,
+    xAxisKeys.length === 1 &&
-        ticks: allTicksForScatter
+    datasetOptions.ticks[0]?.length === 1 &&
-      }
+    xColumnLabelFormat.columnType === 'date' &&
-    };
+    xColumnLabelFormat.style === 'date';
 
-    return lineSeriesBuilder_labels(modifyFiedProps);
+  if (useDateLabels) {
+    // Process date labels directly without extra iterations
+    return datasetOptions.ticks.flatMap((item) =>
+      item.map<Date>((dateItem) => createDayjsDate(dateItem as string).toDate())
+    );
   }
 
-  return undefined;
+  // Only process relevant datasets
+  const relevantDatasets = datasetOptions.datasets.filter((dataset) =>
+    relevantYAxisKeys.has(dataset.dataKey)
+  );
+
+  // Early return for no relevant datasets
+  if (!relevantDatasets.length) return undefined;
+
+  // Collect all ticks without deduplication
+  const allTicks: (string | number)[][] = [];
+  relevantDatasets.forEach((dataset) => {
+    dataset.ticksForScatter?.forEach((tick) => {
+      allTicks.push(tick);
+    });
+  });
+
+  // Sort and flatten
+  return allTicks.sort((a, b) => (a[0] < b[0] ? -1 : a[0] > b[0] ? 1 : 0)).flat();
 };

web/src/components/ui/charts/chartHooks/useDatasetOptions/useDataTrendlineOptions/trendlineDatasetCreator.ts

@@ -1,7 +1,7 @@
 // Also consider modifying this package to make it work with chartjs 4 https://pomgui.github.io/chartjs-plugin-regression/demo/
 
 import type { BusterChartProps, Trendline } from '@/api/asset_interfaces/metric/charts';
-import type { DatasetOption, DatasetOptionsWithTicks } from '../interfaces';
+import type { DatasetOptionsWithTicks } from '../interfaces';
 import type { TrendlineDataset } from './trendlineDataset.types';
 import { DATASET_IDS } from '../config';
 import { isDateColumnType, isNumericColumnType } from '@/lib/messages';
@@ -230,9 +230,6 @@ export const trendlineDatasetCreator: Record<
 
     if (!selectedDatasets || selectedDatasets.length === 0 || validData.length === 0) return [];
 
-    const isXAxisNumeric = isNumericColumnType(
-      columnLabelFormats[xAxisColumn]?.columnType || DEFAULT_COLUMN_LABEL_FORMAT.columnType
-    );
     const isXAxisDate = isDateColumnType(columnLabelFormats[xAxisColumn]?.columnType);
 
     // Get mapped data points using the updated dataMapper