vb6hobbyst7/ssci

JavaScript smoothing, seasonal and regression functions

ssci

JavaScript smoothing, seasonal and regression functions

This library contains functions to smooth, deseasonalise and forecast data series.

It was originally written to help with modifying data prior to charting it via the D3 JavaScript library but can be used more generally (and with other charting libraries).

As the focus of the library is on time series data, most of the functions require a 2 dimensional set of data (i.e. a set of points with x and y coordinates). The data can be in any form and is accessed by functions passed to the function. So data is generally passed via a: .data(data) function. And x coordinates are specified by passing a function via: .x(function(d){ return d.year }) In this example the x data is in the year variable within the data object. A similar function exists for the y coordinate.

The default data accessor functions treat the data as an array of arrays of x and y corrdinates: [ [x1, y1], [x2, y2] , [x3,y3] ... ]

Download

Use the ssci.js or ssci.min.js files from the distrib folder.

Dependencies

This library relies on big.js. This is used as overflow errors occur otherwise when fitting polynomials. The functions that are definitely affected are smooth.quadraticBig, reg.polyBig and reg.determinantBig.

I've also changed the DP variable within this library to 120 as 20 decimal places is not enough to stop odd results being generated via the quadratic smoothing algorithm.

Usage

Add the following tags to your HTML:

<script src="your-folder/big.js"></script>
<script src="your-folder/ssci.js"></script>

Documentation

See ssci for more details and a more detailed description of the functions.

Functions

Smoothing Functions

• Kernel smoothing
• Quadratic smoothing
• Filter smoothing
• Exponentially weighted moving average

Seasonality Functions

• Deseasonalise by average for the period
• Deseasonalise by seasonal difference
• Deseasonalise by moving average

Regression Functions

• Polynomial least squares regression

Forecasting functions

• Exponential smoothing
• Holt exponential smoothing
• Holt Winters exponential smoothing

Time Series Functions

• Auto correlation
• Partial auto correlation
• Difference

Utility Functions

• Change array of objects to array of points
• Change array of points to array of objects
• Create layer for stack layout
• Create layers for stack layout
• Produce cubic interpolation string
• Calculate determinant
• Gamma function

To do

• Tests
• Some of the JSDoc comments need updating to reflect the changes in 1.2