MyFitParamScaling.m
b69aadaa0e93TekRSA-Freq-vs-Time

MyFitParamScaling.m
View Options

			% Class that adds the capability of normalizing the data by z-score before
			% performing the fit to improve numeric performance.
			% Scaling/unscaling functions for the parameters must be defined in
			% the end classes.

			classdef (Abstract) MyFitParamScaling < MyFit

			methods (Access = public)
			function this = MyFitParamScaling(varargin)
			this@MyFit(varargin{:});
			end
			end

			methods (Access = protected)

			% Overload the doFit function to fit scaled data.
			function fitted_vals = doFit(this, x, y, init_vals, lim_lower, ...
			lim_upper)

			% Scale x and y data
			mean_x = mean(x);
			std_x = std(x);
			scaled_x = (x - mean_x)/std_x;

			mean_y = mean(y);
			std_y = std(y);
			scaled_y = (y - mean_y)/std_y;

			% Vector of scaling coefficients
			sc = {mean_x, std_x, mean_y, std_y};

			scaled_fitted_vals = doFit@MyFit(this, scaled_x, scaled_y, ...
			scaleFitParams(this, init_vals, sc), ...
			scaleFitParams(this, lim_lower, sc), ...
			scaleFitParams(this, lim_upper, sc));

			fitted_vals = unscaleFitParams(this, scaled_fitted_vals, sc);
			end
			end

			methods (Access = protected, Abstract)

			% Functions that define scaling and unscaling of the fit parameters
			% by zscore, the zscore is a vector {mean_x, std_x, mean_y, std_y}
			sc_vals = scaleFitParams(~, vals, xy_zscore)

			vals = unscaleFitParams(~, sc_vals, xy_zscore)
			end
			end