refactor(lib): Liberate splitTestData and improve its tests

Author: isair

src/loadCsv.ts

@@ -5,27 +5,10 @@ import { shuffle } from 'shuffle-seed';
 
 import { CsvReadOptions, CsvTable } from './loadCsv.models';
 import filterColumns from './filterColumns';
+import splitTestData from './splitTestData';
 
 const defaultShuffleSeed = 'mncv9340ur';
 
-const splitTestData = (
-  features: CsvTable,
-  labels: CsvTable,
-  splitTest: boolean | number
-) => {
-  const length =
-    typeof splitTest === 'number'
-      ? Math.max(0, Math.min(splitTest, features.length - 1))
-      : Math.floor(features.length / 2);
-
-  return {
-    testFeatures: features.slice(length),
-    testLabels: labels.slice(length),
-    features: features.slice(0, length),
-    labels: labels.slice(0, length),
-  };
-};
-
 const loadCsv = (filename: string, options: CsvReadOptions) => {
   const {
     featureColumns,

src/splitTestData.ts

@@ -0,0 +1,21 @@
+import { CsvTable } from './loadCsv.models';
+
+const splitTestData = (
+  features: CsvTable,
+  labels: CsvTable,
+  splitTest: true | number
+) => {
+  const length =
+    typeof splitTest === 'number'
+      ? Math.max(0, Math.min(splitTest, features.length - 1))
+      : Math.floor(features.length / 2);
+
+  return {
+    testFeatures: features.slice(length),
+    testLabels: labels.slice(length),
+    features: features.slice(0, length),
+    labels: labels.slice(0, length),
+  };
+};
+
+export default splitTestData;

tests/splitTestData.test.ts

@@ -0,0 +1,81 @@
+import splitTestData from '../src/splitTestData';
+
+const tables = {
+  features: [
+    [1, 2],
+    [3, 4],
+    [5, 6],
+    [7, 8],
+  ],
+  labels: [[9], [10], [11], [12]],
+};
+
+test('Default splitting, splits in half', () => {
+  const { features, labels, testFeatures, testLabels } = splitTestData(
+    tables.features,
+    tables.labels,
+    true
+  );
+  expect(features).toMatchObject([
+    [1, 2],
+    [3, 4],
+  ]);
+  expect(labels).toMatchObject([[9], [10]]);
+  expect(testFeatures).toMatchObject([
+    [5, 6],
+    [7, 8],
+  ]);
+  expect(testLabels).toMatchObject([[11], [12]]);
+});
+
+test('Splitting a fixed amount works', () => {
+  const { features, labels, testFeatures, testLabels } = splitTestData(
+    tables.features,
+    tables.labels,
+    1
+  );
+  expect(features).toMatchObject([
+    [1, 2],
+    [3, 4],
+    [5, 6],
+  ]);
+  expect(labels).toMatchObject([[9], [10], [11]]);
+  expect(testFeatures).toMatchObject([[7, 8]]);
+  expect(testLabels).toMatchObject([[12]]);
+});
+
+test('Splitting more than row length splits all rows into test data', () => {
+  const { features, labels, testFeatures, testLabels } = splitTestData(
+    tables.features,
+    tables.labels,
+    tables.features.length * 2
+  );
+  expect(features).toMatchObject([]);
+  expect(labels).toMatchObject([]);
+  expect(testFeatures).toMatchObject([
+    [1, 2],
+    [3, 4],
+    [5, 6],
+    [7, 8],
+  ]);
+  expect(testLabels).toMatchObject([[9], [10], [11], [12]]);
+});
+
+test('Splitting less than or equal to 0 places all rows into normal data', () => {
+  [0, -1].forEach((splitLength) => {
+    const { features, labels, testFeatures, testLabels } = splitTestData(
+      tables.features,
+      tables.labels,
+      splitLength
+    );
+    expect(features).toMatchObject([
+      [1, 2],
+      [3, 4],
+      [5, 6],
+      [7, 8],
+    ]);
+    expect(labels).toMatchObject([[9], [10], [11], [12]]);
+    expect(testFeatures).toMatchObject([]);
+    expect(testLabels).toMatchObject([]);
+  });
+});