---

nonstat.cpp File Reference

Wavelet spectral analysis for non-stationary data (e.g., equity close prices). More...

#include <assert.h>
#include <stdio.h>
#include <vector>
#include "signalUtil.h"
#include "spectrum.h"
#include "yahooTS.h"
#include "line.h"

Go to the source code of this file.

Functions
void	calcReturn (double *ret, const double *close, size_t N)
int	main ()
	\function. More...

---

Detailed Description

Wavelet spectral analysis for non-stationary data (e.g., equity close prices).

Non-stationary signals are signals that do repeat infinitely, as sine and cosine based signals do.

The documentation in this file is formatted for doxygen (see www.doxygen.org).

Copyright and Use

You may use this source code without limitation and without fee as long as you include: This software was written and is copyrighted by Ian Kaplan, Bear Products International, www.bearcave.com, 2002.

This software is provided "as is", without any warranty or claim as to its usefulness. Anyone who uses this source code uses it at their own risk. Nor is any support provided by Ian Kaplan and Bear Products International.

Please send any bug fixes or suggested source changes to:

     iank@bearcave.com

Author:

Ian Kaplan

Definition in file nonstat.cpp.

---

Function Documentation

void calcReturn ( double * ret, const double * close, size_t N )

Definition at line 51 of file nonstat.cpp. Referenced by main(). { for (size_t i = 1; i <= N; i++) ret[i-1] = (close[i] - close[i-1])/close[i-1]; }

int main ( )

\function. The entry point for code to test the wavelet packet transform. The code in main provides a simple example of how to call the wavelet packet transform code. This test case can be used with various wavelet functions (e.g., Haar, linear interpolation, Daubechies D4, polynomial interpolation). In the context of filters, the polynomial interpolation wavelets (see poly.h and polyHaar.h) are very bad choices. The polynomial interpolation wavelets do not divide the spectrum into high and low frequency components. Instead, the power spectrum shows one peak. I'm not sure if this means that the polynomial interpolation wavelets are bad for all applications. For example, another way to look at wavelets is as approximation functions. For example, wavelet compression relies approximation functions. In this case polynomial wavelets may be a good choice for some data sets. Definition at line 82 of file nonstat.cpp. { const size_t NClose = 513; double close[NClose]; const char *path = "..\\data\\equities\\"; yahooTS ts( path ); const char *stock = "amat"; size_t n = NClose; ts.getTS( stock, close, n, yahooTS::Close ); if (n != NClose) { printf("Error reading file %s%s\n", path, stock ); return 0; } const size_t N = 512; double ret[ N ]; calcReturn( ret, close, N ); // prVec( ret, N ); // linear interpolation wavelet line<double *> w; double *ptr = ret; w.forwardTrans( ptr, N ); std::vector<double> powerVec; spectrum::spectralCalc( ret, N, powerVec ); // // Print out the power spectrum // size_t len = powerVec.size(); for (size_t i = 0; i < len; i++) { // printf("%2d, %7.4f\n", i, powerVec[i] ); } double d1[N], d2[N]; spectrum::copyBands( d1, ret, N, 0, 7 ); spectrum::copyBands( d2, ret, N, 8, 9 ); ptr = d1; w.inverseTrans( ptr, N ); ptr = d2; w.inverseTrans( ptr, N ); // // Print out the lower half of the spectrum // // signalUtil::prVec( d1, N ); // // Print out the upper half of the spectrum // signalUtil::prVec( d2, N ); return 0; }

---