iyidan/analytics

Data analytics and curve fitting in go

analytics

Analytics and curve fitting library in go.
The library provides a Series structure that can be initialized with two []float64 containing x/y data.
Once initialized, functions can be applied to create new Series, or curve fitting can be applied to interpolate or (depending on the algorithm) extrapolate data points.

Simple example usage:

	//Define a small dataset
	x := []float64{1, 2, 3, 4, 5, 6, 7, 8}
	y := []float64{8, 6, 5, 5, 4, 3, 3, 2}
	
	//Use the dataset as a series
	series1 := analytics.NewSeriesFrom(x, y)

	//Get a 3rd Order Polynomial fit for the series
	fit := series1.FitPolynomial(3)

	//Display two interpolated/extrapolated points
	fmt.Println(analytics.Extrapolate(fit, 4))
	fmt.Println(analytics.Extrapolate(fit, 9))

	//Create smoothed version of the dataset
	series2 := series1.Smoother(3)

	//Display the underlying smoothed values
	x1, y1 := series2.ToArrays()
	fmt.Println(series2.x1)
	fmt.Println(series2.y1)

##General Data Manipulation Functions
Smoother - Iterative noise removal algorithm.
Pixelize - Quantization function.
MA - Moving average
EMA - Exponential moving average
LWMA - Linear weighted moving average
TrendChanges - Apex for peaks and troughs for smoothed data.
ApplyOffset - Move a series on x/y axes

##Data Splicing and Combining Functions
RecentTrends - Splices smoothed data into multiple series, each describing a trend.
Last - Extracts a copy of the last n points from the end of a series.
From - Extracts a copy of data points starting from an arbitrary x value.
Append - Joins two series together to form a new series.

##Financial Analysis Based Functions
ITrend - John Ehlers instantaneous trend (iTrend) indicator
CCI - Commodity Channel Index

##Misc Functions ToArrays - Extracts two 1D slices of values, one for x and one for y ToValues - As ToArrays, but takes an offset from the last datapoint

##Curve fit types Linear
Logarithmic
Exponential
Power
Polynomial (n-order)
Gaussian
Parabolic

##Todo Tests! (started) Fix up Gaussian fit (use scaling in addition to the offset) Fix up Parabolic fit (use scaling in addition to the offset)
Implement Cubic Spline
Cache sums for curve fit
Error handling