getVariance1()

src/main/ml.ts

@@ -79,56 +79,4 @@ export const mlStandardizeTestData = (data: number[], decimalPlaces = Infinity):
  */
 export const mlStandardizeUnseenData = (data: number[], mean: number, stdDev: number, decimalPlaces = Infinity): number[] => {
     return mlStandardizeArray(data, mean, stdDev, decimalPlaces);
-};
-
-/*
-// which means that 1 in 5 patients will miss their scheduled appointment.
-def calc_prevalence(y):
- return (sum(y)/len(y))
- */
-
-/*
-df[‘ScheduledDay_year’] = df[‘ScheduledDay’].dt.year
-df[‘ScheduledDay_month’] = df[‘ScheduledDay’].dt.month
-df[‘ScheduledDay_week’] = df[‘ScheduledDay’].dt.week
-df[‘ScheduledDay_day’] = df[‘ScheduledDay’].dt.day
-df[‘ScheduledDay_hour’] = df[‘ScheduledDay’].dt.hour
-df[‘ScheduledDay_minute’] = df[‘ScheduledDay’].dt.minute
-df[‘ScheduledDay_dayofweek’] = df[‘ScheduledDay’].dt.dayofweek
- */
-
-// I would never use Year as a feature (but showed it anyways) because presumably we want to use this predictive model in the future and those future years wouldn’t be included in the dataset.
-
-// However, I am kind of disappointed that the months are just for a few months of the year. This means month (and consequently week of year) probably shouldn’t be used as a feature either.
-
-/*
-// the number of days between the scheduled date and the appointment date.
-df[‘delta_days’] = (df[‘AppointmentDay’]-df[‘ScheduledDay’]).dt.total_seconds()/(60*60*24)
- */
-
-// Another thing you may want to do is convert the dayofweek into a categorical variable via one-hot encoding.
-
-// Which shows that delta_days is basically the only features used in the model. This confirms our suspicion above that the model will likely struggle due to those same day appointments.
-
-/*
-When working with months (and days) as integers you are losing some information about them, because December (12) is far away from January (1) but in a sense they should be connected as one precedes the other. One approach to solve this problem would be projecting the months into a circle, then each month would have a corresponding angle “theta” and then you can store the “sin” and “cos” of said angle.
- */
-
-/*
-Possible suggestions of features include:
-
-weekends vs weekdays
-business hours and time of day
-seasons
-week of year number
-month
-year
-beginning/end of month (pay days)
-quarter
-days to/from an action event(distance)
-missing or incomplete data
-etc.
-All this depends on the data set and most won't apply.
-
-
- */
+};

src/main/statistics.ts

@@ -77,6 +77,7 @@ TODO:
 
 /**
  * Dispersion: the average square distance from the mean.
+ * Sum of (x - mean)^2 / N
  */
 export const getVariance = (data: number[], decimalPlaces = Infinity) : number|undefined => {
     if(!data || data.length <= 0) return undefined;
@@ -89,6 +90,21 @@ export const getVariance = (data: number[], decimalPlaces = Infinity) : number|u
     return setDecimalPlaces(sum / data.length, decimalPlaces);
 };
 
+/**
+ * Another formula
+ * (Sum of x^2) / N - (mean ^ 2)
+ */
+export const getVariance1 = (data: number[], decimalPlaces = Infinity) : number|undefined => {
+    if(!data || data.length <= 0) return undefined;
+
+    const mean = getArithmeticMean(data);
+    if(mean === undefined) return undefined;
+
+    const sum = data.reduce((acc, val) => acc + (val ** 2), 0);
+
+    return setDecimalPlaces((sum / data.length) - (mean ** 2), decimalPlaces);
+};
+
 export const getStandardDeviation = (data: number[], decimalPlaces = Infinity) => {
     const variance = getVariance(data) ?? 0;
     return setDecimalPlaces(Math.sqrt(variance), decimalPlaces);

test/statistics.test.ts

@@ -1,4 +1,11 @@
-import { getArithmeticMean, getMedian, getMode, getVariance, getStandardDeviation } from '../src/main/statistics';
+import {
+    getArithmeticMean,
+    getMedian,
+    getMode,
+    getVariance,
+    getVariance1,
+    getStandardDeviation
+} from '../src/main/statistics';
 
 describe('Statistics', () => {
 
@@ -104,6 +111,29 @@ describe('Statistics', () => {
             });
         });
 
+        describe('getVariance1()', () => {
+
+            test('[]', () => {
+                expect(getVariance1([])).toStrictEqual(undefined);
+            });
+
+            test('[2]', () => {
+                expect(getVariance1([2])).toStrictEqual(0);
+            });
+
+            test('[2, 2, 3, 3]', () => {
+                expect(getVariance1([2, 2, 3, 3])).toStrictEqual(0.25);
+            });
+
+            test('[0, 0, 5, 5]', () => {
+                expect(getVariance1([0, 0, 5, 5])).toStrictEqual(6.25);
+            });
+
+            test('[1, 2, 3, 8, 7]', () => {
+                expect(getVariance1([1, 2, 3, 8, 7], 2)).toStrictEqual(7.76);
+            });
+        });
+
         describe('getStandardDeviation()', () => {
 
             it('returns 0 when data is empty', () => {