feat: add xBoxPlotWithOutliers mehtod to deal with possible outliers (#…

…260) This removes the previous option in xBoxPlot that would have yield to different types
mljs · Oct 1, 2024 · 985caee · 985caee
1 parent 53e46f0
commit 985caee
Show file tree

Hide file tree

Showing 6 changed files with 162 additions and 45 deletions.
diff --git a/src/__tests__/__snapshots__/index.test.ts.snap b/src/__tests__/__snapshots__/index.test.ts.snap
@@ -14,6 +14,7 @@ exports[`test existence of exported functions 1`] = `
   "xApplyFunctionStr",
   "xAutoCorrelation",
   "xBoxPlot",
+  "xBoxPlotWithOutliers",
   "xCheck",
   "xCheckLengths",
   "xCorrelation",

diff --git a/src/x/__tests__/xBoxPlot.test.ts b/src/x/__tests__/xBoxPlot.test.ts
@@ -10,7 +10,6 @@ test('test xBoxPlot even', () => {
     q3: 8.5,
     min: 0,
     max: 11,
-    outliers: [],
   });
 });
 
@@ -22,7 +21,6 @@ test('test xBoxPlot even small', () => {
     q3: 4,
     min: 0,
     max: 5,
-    outliers: [],
   });
 });
 
@@ -34,7 +32,6 @@ test('test xBoxPlot odd', () => {
     q3: 8,
     min: 0,
     max: 10,
-    outliers: [],
   });
 });
 
@@ -46,7 +43,6 @@ test('test xBoxPlot odd small', () => {
     q3: 3.5,
     min: 0,
     max: 4,
-    outliers: [],
   });
 });
 
@@ -73,7 +69,6 @@ test('test xBoxPlot with one element', () => {
     q3: 42,
     min: 42,
     max: 42,
-    outliers: [],
   });
 });
 
@@ -86,7 +81,6 @@ test('test xBoxPlot with 2 elements', () => {
     q3: 44,
     min: 40,
     max: 44,
-    outliers: [],
   });
 });
 
@@ -99,7 +93,6 @@ test('test xBoxPlot with 3 elements', () => {
     q3: 44,
     min: 40,
     max: 44,
-    outliers: [],
   });
 });
 
@@ -111,18 +104,5 @@ test('outliers', () => {
     q3: 8,
     min: 0,
     max: 100,
-    outliers: [],
-  });
-});
-
-test('outliers', () => {
-  const array = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 100];
-  expect(xBoxPlot(array, { calculateOutliers: true })).toStrictEqual({
-    q1: 2,
-    median: 5,
-    q3: 8,
-    min: 0,
-    max: 9,
-    outliers: [100],
   });
 });
diff --git a/src/x/__tests__/xBoxPlotWithOutliers.test.ts b/src/x/__tests__/xBoxPlotWithOutliers.test.ts
@@ -0,0 +1,116 @@
+import { expect, test } from 'vitest';
+
+import { xBoxPlotWithOutliers } from '../xBoxPlotWithOutliers';
+
+test('test xBoxPlotWithOutliers even', () => {
+  const array = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11];
+  expect(xBoxPlotWithOutliers(array)).toStrictEqual({
+    q1: 2.5,
+    median: 5.5,
+    q3: 8.5,
+    min: 0,
+    max: 11,
+    outliers: [],
+  });
+});
+
+test('test xBoxPlotWithOutliers even small', () => {
+  const array = [0, 1, 2, 3, 4, 5];
+  expect(xBoxPlotWithOutliers(array)).toStrictEqual({
+    q1: 1,
+    median: 2.5,
+    q3: 4,
+    min: 0,
+    max: 5,
+    outliers: [],
+  });
+});
+
+test('test xBoxPlotWithOutliers odd', () => {
+  const array = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10];
+  expect(xBoxPlotWithOutliers(array)).toStrictEqual({
+    q1: 2,
+    median: 5,
+    q3: 8,
+    min: 0,
+    max: 10,
+    outliers: [],
+  });
+});
+
+test('test xBoxPlotWithOutliers odd small', () => {
+  const array = [0, 1, 2, 3, 4];
+  expect(xBoxPlotWithOutliers(array)).toStrictEqual({
+    q1: 0.5,
+    median: 2,
+    q3: 3.5,
+    min: 0,
+    max: 4,
+    outliers: [],
+  });
+});
+
+test('test xBoxPlotWithOutliers too small', () => {
+  const array = [0, 1, 2, 3];
+  expect(() => xBoxPlotWithOutliers(array)).toThrow(
+    'can not calculate info if array contains less than 5 elements',
+  );
+});
+
+test('test xBoxPlotWithOutliers with one element', () => {
+  const array = [42];
+  expect(() => xBoxPlotWithOutliers(array)).toThrow(
+    'can not calculate info if array contains less than 5 elements',
+  );
+});
+
+test('test xBoxPlotWithOutliers with one element', () => {
+  const array = [42];
+  const result = xBoxPlotWithOutliers(array, { allowSmallArray: true });
+  expect(result).toStrictEqual({
+    q1: 42,
+    median: 42,
+    q3: 42,
+    min: 42,
+    max: 42,
+    outliers: [],
+  });
+});
+
+test('test xBoxPlotWithOutliers with 2 elements', () => {
+  const array = [40, 44];
+  const result = xBoxPlotWithOutliers(array, { allowSmallArray: true });
+  expect(result).toStrictEqual({
+    q1: 40,
+    median: 42,
+    q3: 44,
+    min: 40,
+    max: 44,
+    outliers: [],
+  });
+});
+
+test('test xBoxPlotWithOutliers with 3 elements', () => {
+  const array = [40, 42, 44];
+  const result = xBoxPlotWithOutliers(array, { allowSmallArray: true });
+  expect(result).toStrictEqual({
+    q1: 40,
+    median: 42,
+    q3: 44,
+    min: 40,
+    max: 44,
+    outliers: [],
+  });
+});
+
+test('outliers', () => {
+  const array = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 100];
+  expect(xBoxPlotWithOutliers(array)).toStrictEqual({
+    q1: 2,
+    median: 5,
+    q3: 8,
+    min: 0,
+    max: 9,
+    outliers: [100],
+  });
+});
diff --git a/src/x/index.ts b/src/x/index.ts
@@ -6,6 +6,7 @@ export * from './xAdd';
 export * from './xApplyFunctionStr';
 export * from './xAutoCorrelation';
 export * from './xBoxPlot';
+export * from './xBoxPlotWithOutliers';
 export * from './xCheck';
 export * from './xCheckLengths';
 export * from './xCorrelation';

diff --git a/src/x/xBoxPlot.ts b/src/x/xBoxPlot.ts
@@ -6,12 +6,6 @@ export interface XBoxPlotOptions {
    * @default false
    */
   allowSmallArray?: boolean;
-
-  /**
-   * Calculate outliers (value < min-1.5IQR or value > max+1.5IQR). The min and max are recalculated without the outliers.
-   * @default false
-   */
-  calculateOutliers?: boolean;
 }
 
 export interface XBoxPlot {
@@ -20,7 +14,6 @@ export interface XBoxPlot {
   q3: number;
   min: number;
   max: number;
-  outliers: number[];
 }
 
 /**
@@ -32,7 +25,7 @@ export function xBoxPlot(
   array: NumberArray,
   options: XBoxPlotOptions = {},
 ): XBoxPlot {
-  const { allowSmallArray = false, calculateOutliers = false } = options;
+  const { allowSmallArray = false } = options;
   if (array.length < 5) {
     if (allowSmallArray) {
       if (array.length === 0) {
@@ -53,7 +46,6 @@ export function xBoxPlot(
     q3: 0,
     min: array[0],
     max: array.at(-1) as number,
-    outliers: [],
   };
   let q1max, q3min;
   if (array.length % 2 === 1) {
@@ -77,21 +69,5 @@ export function xBoxPlot(
     info.q3 = (array[middleOver] + array[middleOver - 1]) / 2;
   }
 
-  if (calculateOutliers) {
-    const iqr = info.q3 - info.q1;
-    const min = info.q1 - 1.5 * iqr;
-    const max = info.q3 + 1.5 * iqr;
-    // we need to recalculate the min and the max because they could be outliers
-    info.min = info.median;
-    info.max = info.median;
-    for (const value of array) {
-      if (value < min || value > max) {
-        info.outliers.push(value);
-      } else {
-        if (value < info.min) info.min = value;
-        if (value > info.max) info.max = value;
-      }
-    }
-  }
   return info;
 }
diff --git a/src/x/xBoxPlotWithOutliers.ts b/src/x/xBoxPlotWithOutliers.ts
@@ -0,0 +1,43 @@
+import { NumberArray } from 'cheminfo-types';
+
+import { xBoxPlot, XBoxPlotOptions } from './xBoxPlot';
+
+export interface XBoxPlotWithOutliers {
+  q1: number;
+  median: number;
+  q3: number;
+  min: number;
+  max: number;
+  outliers: number[];
+}
+
+/**
+ * Calculating the box plot of the array
+ * @param array - data
+ * @param options
+ */
+export function xBoxPlotWithOutliers(
+  array: NumberArray,
+  options: XBoxPlotOptions = {},
+): XBoxPlotWithOutliers {
+  const info: XBoxPlotWithOutliers = {
+    ...xBoxPlot(array, options),
+    outliers: [],
+  };
+
+  const iqr = info.q3 - info.q1;
+  const min = info.q1 - 1.5 * iqr;
+  const max = info.q3 + 1.5 * iqr;
+  // we need to recalculate the min and the max because they could be outliers
+  info.min = info.median;
+  info.max = info.median;
+  for (const value of array) {
+    if (value < min || value > max) {
+      info.outliers.push(value);
+    } else {
+      if (value < info.min) info.min = value;
+      if (value > info.max) info.max = value;
+    }
+  }
+  return info;
+}