The SDL Component Suite is an industry leading collection of components supporting scientific and engineering computing. Please visit the SDL Web site for more information....


The class TCurveFit provides a simple-to-use means for curve fitting by means of linear regression and splines. In addition, TCurveFit calculates the most important statistical parameters, such as the mean values, the standard deviation, and the correlation coefficient of a series of data pairs. In order to utilize the class TCurveFit, you have to enter the data pairs (x and y) by the method EnterStatValue. The regression parameters can then be obtained by calling the appropriate regression method.

The following table summarizes the routines of TCurveFit involved in curve fitting:

Init initialize the processing machine
EnterStatValue enter a pair of data
RemoveStatValue remove a data pair from the pool
DataX, DataY direct (read-only) access to the entered data pairs
CalcAnovaReg performs the analysis of variances for linear models
CalcCenteredPolyFit calculate the best fitting centered polynomial
CalcCircleFit calculate the best fitting circle
CalcExponentialFit calculate the best exponential fit
CalcGaussFit calculate the best Gaussian fit (normal distribution)
CalcHoerlFit calculate the best fitting Hoerl function
CalcHyperbolFit calculate the best hyperbolic fit to the data
CalcLinFit calculate the best linear fit to the data
CalcLogFit calculate the best logarithmic fit
CalcParabolFit calculate the best parabolic fit to the data
CalcPolyFit calculate a best fitting polynomial
CalcPowerFit calculate a best fitting power curve
CalcReciHyperbolFit calculate the best reciprocal hyperbolic fit to the data
CalcReciLinFit calculate the best reciprocal linear fit to the data
CalcReciLogFit calculate the best reciprocal logarithmic fit
CalcRegModel calculate a regression model
CPXShift the shift along the x axis of centered polynomials
CubicSpline calculates a cubic spline interpolation
Evaluate evaluates the regression model at a specific x value
FitQual the quality of fit of the previously calculated model
PolyOrder order of the previously calculated polynomial regression
Predict predict a value at a particular x value
Residual the residuals of the regression
StdDevResid the standard deviation of the residuals of the regression
SmoothedSpline calculates a smoothed cubic spline interpolation
SpearmanRankCorr calculates Spearman's rank correlation coefficient

In addition, TCurveFit also provides all important univariate and bivariate parameters for the entered data pairs:

CalcStatistics calculate mean values, standard deviations, and the correlation coefficient
CorrCoeff correlation coefficient
KendallsTau Kendalls's Tau (rank correlation coefficient)
KruskalGamma Kruskal's Gamma (rank correlation coefficient)
KurtosisX, KurtosisY kurtosis of x and y values
MaxX, MaxY maxima of x and y values
MeanDiff mean of differences between x and y
MeanAngleX mean of x values interpreted as angles measured in degrees
MeanX mean of x values
MeanAngleY mean of y values interpreted as angles measured in degrees
MeanY mean of y values
MinX, MinY minima of x and y values
NumData number of data pairs
RMSDiff root mean square of the differences between x and y
SkewnessX, SkewnessY skewness of x and y values
StdDevDiff standard deviation of differences between x and y
StdDevX standard deviation of x values
StdDevY standard deviation of y values

Last Update: 2023-Dec-11