hidalgo-cusum-test-stat

Hidalgo.js library to calculate CUSUM-based test statistics

Yes it's a long name, but at least it's descriptive.

Install

$ npm install --save hidalgo-cusum-test-stat

Usage

For best results, use with hidalgo-cusum-pvalue:

var testStat = require('hidalgo-cusum-test-stat');
var pValue = require('hidalgo-cusum-pvalue');
 
var stats = testStat([0, 1, 2, 3 /*, ... more data ... */], 28, 2);
var pValues = stats.map(pValue);

API

cusum(data, baseline, guardBand [, options])

Calculates an array of test statistics. Returns an array of numbers of length data.length - baseline - guardBand.

This function takes the following mandatory parameters:

• data - array of numbers representing data points
• baseline - number of previous data points to use. This should be at least 28 for best results.
• guardBand - number of data points in between the baseline and the point to calculate

Formally, cusum[i] is a function of data[i - baseline - guardBand..i - guardBand]. In other words, if your baseline is 28 and your guard band is 2 (which are the recommended defaults), then CUSUM will look at the 28 data points preceding i-2 to calculate cusum[i]. This means that CUSUM will not output values for the first baseline + guardBand values in data.

cusum also takes an optional options parameter. options is an object with the following keys:

• generateBaseline - Function to generate baseline data from the input array. Use this to alter the data set for example, by removing long strings of 0 counts. Defaults to returning the input array unchanged
• cusumK - Number to subtract from the test stat. Defaults to 0.5
• minSigma - Minimum σ. The standard deviation of the baseline data is used only if it is greater than minSigma. Otherwise, minSigma is used. Defaults to 0.5
• resetLevel - Test statistics larger than resetLevel are reduced. Defaults to 4