Levenshtein distance

src/widgets/MetricsWidget/levenshtein.ts

@@ -0,0 +1,16 @@
+export function computeLevenshtein(actualValues: string, assignedValues: string): number {
+    if (actualValues.length === 0) {
+        return assignedValues.length;
+    }
+    if (assignedValues.length === 0) {
+        return actualValues.length;
+    }
+
+    const cost = assignedValues[0] === actualValues[0] ? 0 : 1;
+
+    const deletion = computeLevenshtein(actualValues.slice(1), assignedValues) + 1;
+    const insertion = computeLevenshtein(actualValues, assignedValues.slice(1)) + 1;
+    const substitution = computeLevenshtein(actualValues.slice(1), assignedValues.slice(1)) + cost;
+
+    return Math.min(deletion, insertion, substitution);
+}

src/widgets/MetricsWidget/metrics.ts

@@ -1,6 +1,7 @@
 import _ from 'lodash';
 import { Metric } from './types';
 import { computeConfusion } from './confusion';
+import { computeLevenshtein } from './levenshtein';
 
 export const METRICS: Record<string, Metric> = {
     sum: {
@@ -97,4 +98,22 @@
             );
         },
     },
+    Levenshtein: {
+        signature: {
+            X: ['str'],
+            Y: ['str'],
+        },
+        compute: ([actualValues, assignedValues]) => {
+            const numValues = actualValues.length;
+            var sumLevenshtein = 0;
+
+            for (let i = 0; i < actualValues.length; i++) {
+                const actual = actualValues[i];
+                const assigned = assignedValues[i];
+                sumLevenshtein += computeLevenshtein(actual as string, assigned as string);
+            }
+
+            return sumLevenshtein / numValues;
+        },
+    },
 };

src/widgets/MetricsWidget/types.ts

@@ -1,6 +1,6 @@
 import { DataKind } from '../../datatypes';
 
-export type ValueArray = number[] | Int32Array | boolean[];
+export type ValueArray = number[] | Int32Array | boolean[] | string[];
 
 export interface Metric {
     signature: Record<string, DataKind | DataKind[]>;