FindGraph is a graphing, curve fitting, and digitizing tool. The program offers 10 generic fits, including linear regression, logistic functions, rational, Fourier approximation, neural networks, plus a library of over 500 industry-specific 2D formulas. Built-in Wizard helps applying different curve fits to discover the model that describes data the best. There is a feature to select best fit function based on minimum least squares value, or BDS test value, or minimum BIC (Bayesian information criterion) value. FindGraph enables to smooth, interpolate, subtract, differentiate, integrate, and transform data and curves, to apply the Douglas-Peucker algorithm for reducing the number of points, to apply FFT and wavelet filters, Epanechnikov kernel smoother, to estimate Hurst or Lyapunov exponent, to apply SSA, ARIMA, and to extract cyclic signals. It includes feature to apply GMDH Polynomial Neural Network or Radial Basis Function Network to fit and to forecast data series. Procrustes analysis is included. Built-in Wizard of Digitization will help you to digitize the data. Excel examples are included to learn how to call automation methods to fit and to forecast data series.
What is new in this release:
Version 2.59:
- Add feature to find best fit with conditions (for example, second derivative is positive in all dataset)
- Feature to apply regularization (Forward Stagewise Linear Regression algorithm) to prevent overfitting.
What is new in version 2.573:
Version 2.57:
- Feature to apply regularization (Forward Stagewise Linear Regression algorithm) to prevent overfitting.
- Feature to find the decision boundary in logistic regression classification.
What is new in version 2.552:
Version 2.552:
- Epanechnikov quadratic kernel smoother
- Feature to apply boosting (a weighted average of predictions of individual regression lines) in linear regression to prevent overfitting.
Limitations:
30-day trial
Comments not found