Skip to main content

Stats Master Release History

 New features and changes for v6.2.0 (November 2023)

  • Support for Rad Studio Athens 12.0
  • Added ARIMA Trend decomposition example.
  • Bug fix for MannWhitneyTest.
  • Updated for MtxVec v6.2.0

 New features and changes for v6.1.0 (June 2022)

  • Android 11 tagged pointer support.

Time-series forecasting:

  • Rewritten ARMAForecast
  • Rewritten ARMAMLE
  • Added new trend decomposition ARIMA example to Stats Master demo.
  • Added Durbin-Watson statistics.

 New features and changes for v6.0.5 (September 2021)

  • Added support for RAD Studio 11 Alexandria

 New features and changes for v6.0.4 (July 2021)

  • Initial release of the Stats Master for .NET Core 5.0 rewritten in C#.

 New features and changes for v6.0.4 (March 2021)

  • Fixed a bug when fitting parameters of the Fisher random distribution.
  • Vectorized version of the Fisher PDF.

 New features and changes for v6.0.4 (March 2021)

  • Bug fix for Stepwise multiple regression when running with enabled assertions (debug mode).

 New features and changes for v6.0.1 (June 2020)

  • Support for Sydney 10.4 and matching MtxVec v6 features.

 New features and changes for v6.0 (March 2020)

  • Matching MtxVec v6 with features.
  • Added C++Builder support across all features.

 New features and changes for v6.0 (February 2020)

  • Support for Rad Studio XE10.3 Rio Update 3 and Android64 compiler.
  • Support for combined (either double or single) precision where meaningfull.

 New features and changes for v5.3.2 (December 2018)

  • Support for Rad Studio XE10.3 Rio
  • Installer support for FMX and Delphi platforms.

 New features and changes for v5.3.1 (January 2018)

  • New stepwise multiple linear regression component TMtxStepwiseReg. Suport for forward, backward and exhaustive search.
  • Support for Linux in Rad Studio XE10.2 Tokyo (Update 2). The following units are functional in Linux: RegModels, Regress, Statistiscs, StatRandom and StatTools
  • Bug fix for NegBinomPDF and NegBinomCDFInv (numerical stability)
  • .NET bug fix for Marquardt optimization method (giving rise to "complement of two.." type of error).
  • GeometricFit fix for confidence interval estimation.

New features and changes for v5.2 (May 2017)

  • Updated for MtxVec v5.2 (Intel IPP and MKL update).
  • Support for Rad Studio XE10.2 Tokyo.

 New features and changes for v5.1.1 (September 2016)

  • Updated for MtxVec v5.1.1 (Intel IPP and MKL update).
  • Performance improvement for PairwiseDistance function (L2 norm) by roughly 10x.

New features and changes for v5.1 (May 2016)

  • Updated for MtxVec v5.1 (Intel IPP v9 update and Intel MKL update).
  • Cross platform support for FireMonkey on iOS, OS X and Android.
  • Support for old and new C++Builder 32bit and 64bit compilers.
  • Support for XE10.1 and related TeeChart updates.

New features and changes for v5.02 (April 2015)

  • Updated for MtxVec v5.02 (Performance Enhancements, Android support).
  • Update to Stats Master demo for FireMonkey to run also on Android tablets.
  • Support for XE8 and related TeeChart updates.

New features and changes for v5.0 (December 2014)

  • Update for MtxVec v5. All features are now available optionally also with full source pascal code running at lower speed, but with cross-platform portability.
  • Updated for FireMonkey XE7.

New features and changes for v4.4 (September 2014)

  • Added support for XE7.

New features and changes for v4.4 (February 2014)

  • Added support for FireMonkey including all UI controls, Demo and TeeChart support from including XE5 forward.
  • Fixed double and tripple exponential smoothing.
  • Fixed a potential round off error with two-level full factorial design.
  • Added new faster overload for histogram computation.

New features and changes for v4.1 (October 2011)

  • Updated for TeeChart Pro 2011 and Delphi/C++Builder XE2 including support for 64bit compiler.

New features and changes for v4.0 (June 2010):

Statistics.pas, Regress.pas

  • Fixed GrubbsTest
  • Fixed bug with RegressTest and Constant param
  • Fixed bug with TBiPlot demo
  • Updated for TeeChart 2010 and Delphi/C++Builder 2010.

New features and changes for v3.5 (September 2008):

Statistics.pas

  • Support for Delphi 2009 and CBuilder 2009
  • New help files.
  • Hypothesis test - fixed hypothesis result when test statistics was Inf or Nan.
  • Several changes in Covatiance and CorrCoef implementation (vectorized)
  • Fixed several bugs in A-D test. [StatTools.pas]
  • GOFKolmogorov bug fixed.

StatProbPlots.pas

  • Fixed Q-Q plot when plotting only one dataset.

New features and changes for v3.0 (June 2007):

  • Improved random generators.
  • Added Gumbel, Triangular, Erlang distribution random number generators.
  • Changed implementation for all random generators to allow either use of MtxVec.Random.dll or "inhouse"
  • simplified random generators.
  • More examples, including C# an C++ examples.
  • All examples have been updated with the new Vector/Matrix syntax.
  • New HtmlHelp 2 format support. Now F1 works again.
  • Added Chi-Squared distribution parameter estimate (ChiSquareFit).
  • Added Cauchy distribution MLE parameter estimates (CauchyFit).
  • Added Erlang distribution parameter estimates (ErlangFit).
  • Added Laplace distribution MLE parameter estimates (LaplaceFit).
  • Added Negative Binomial distribution parameter estimates (NegBinomFit).
  • Expanded regression routines.
  • Added PRESS and R2 functions.
  • Several bug fixes.

New features and changes for v2.1 (May 2006):

  • Several new routines for handling/manipulating time series.
  • New tests for normal distribution, detecting outliers, etc.
  • Improved examples and help files.
  • Several bug fixes.
  • Added Kolmogorov-Smirnov goodnes of fit test (one and two sample).
  • Added empirical cumulative distribution function.
  • Chi2 GOF test: added support for log-normal and uniform distribution.
  • Added ZScores for vector.
  • Added Bera-Jarque test of composite normality.
  • Added Grubbs's test for outliers.
  • Several bugs in routine Histogram fixed.
  • Added single moving average function.
  • Added sigle, double and triple exponential smoothing routines.
  • Added sigle, double and triple exponential forecasting routines.
  • Added ARMA/ARIMA simulation routines.
  • Added ARMA/ARIMA estimation (MLE) routines.
  • Added ARMA/ARIMA forecasting routines.
  • Added Yule-Walker AR coefficients estimation.
  • Added Burg AR coefficients estimation.
  • Added Innovations ARMA and MA coefficients estimation.
  • Added innovations and Durbin-Levinson algorithms.
  • Added ARAR model fitting and forecasting routines.
  • Added Box-Cox and inverse Box-Cox transformation.
  • Added rational model fit.
  • Added logistic model fit.
  • Added Poisson regression.
  • Added tMtxBinaryTest component for performing one and two test(s) binary diagnostic test.
  • Added several new examples demonstrating time series analysis routines..

New features and changes for v2.0 (December 2005):

  • The demo now compiles with VCL.NET.
  • Fixed a memory leak for the custom editors.
  • Support for Delphi 2005, 2006 W32 and .NET personalities and updated recompile tool.

New features and changes in v2.0 (May 2005):

  • Compatible with MtxVec v2.
  • Support for .NET.
  • Support for SSE3 instruction set.
  • Substantial increase in number of optimized primitive functions and methods via MtxVec v2.
  • Improved range checking and programmer safety.
  • Several bug fixes.
  • Enhanced and more customizable routines.

New features and changes in v1.1 (June 2002):

Statistics.pas

  • Added Chi-Squared and F tests for comparing variance(s).
  • Added Spearman rank correlation test.
  • Added LogNormalFit routine for fitting log-normally distributed values.
  • Fixed the internal TiedRank routine (in some special cases ranks were not calculated correctly) and moved it to public section.
  • Added autocovariance, autocorrelation and partial autocorellation functions.
  • Added Chi2 Goodness of fit test routines.
  • Added Shapiro-Wilks and Shapiro-Francia Goodness of fit test routines.
  • Minor bug fixes.

StatSeries.pas

  • Added control limits calculation for R-Chart.
  • Added EWMA chart calculation.
  • Removed the probability axis tool. Now the relevant code is included in probability series (simplified code - see updated example in Statistics demo).

Regress.pas

  • Fixed R2 calculation for Constant := false case.
  • Added logistic (simple and ordinal) regression routines.
  • Minor bug fixes

StatTools.pas

  • New TMtxLogistReg component which encapsulates logistic regression routines. 

Initial release of Statistics for MtxVec 1.0 (July 2001)

Fixed a potential round off error with two-level full factorial design.

Stats Master - Statistical add-on package for MtxVec

Stats Master

Statistical add-on package for MtxVec

Overview

If your applications requires probability and statistics calculations, our Stats Master statistical software package is just what you need to speed up your application. The languages supported include C#, VB.NET, Delphi and C++Builder.

Stats Master introduces basic and more advanced routines for statistical analysis. Since it's a MtxVec add-on package, it requires the latest MtxVec version installed on your machine.

This package features about 300 different routines (not counting overloads) and several components. The package is built without requiring any additional dlls and full source is available. It takes advantage of major MtxVec features like support for code vectorization, symmetric multiprocessing and a large set of functions already available in MtxVec.

Stats Master for Delphi/C++ Builder

Learn more

Stats Master for Visual Studio .NET

Learn more

Stats Master for MtxVec Features

A complete list of all available routines with examples is available in the help file (Delphi, .NET) and below is a short summary:

  • Distributions

    Different distributions (PDF, CDF and inverse CDF for 36 distributions), random number generators, parameter estimate.

  • Descriptive statistics

    Histograms, ogives, cumulative sum, nth moments, percentile, range, IQR, mean, median, mode, ranks and more.

  • Multivariate Analysis

    PCA by using covariance/correlation matrix, PCA residuals, orthogonal rotation of ZScores, Bartlett test for dimensionality and Z-Scores; Classical Multidimensional Scaling, Hotelling T2 test, M-Box test, Item Analysis.

  • Design of Experiment

    Full Factorial Design, Latin HyperCube design, ...

  • Hypothesis testing

    Sign test, Wilcoxon Signed Rank test, one-sample t-test, two-sample paired/unpaired t-test, Z-Test, Chi-Squared test, F-Test, Shapiro-Wilks test, Chi-Squared Goodness of Fit test and Shapiro-Francia test; Berra-Jarque, Anderson-Darling, Kolmogorov-Smirnov test, Mann-Whitney U test, LillioeFors Goodness of Fit test, ...

  • Regression models

    Linear (weighted, unweighted), Multiple linear (weighted, unweighted), Logistic regression, Ridge regression, Poisson regression, General non-linear regression (using the BFGS, Marquardt, Conjugate gradient or Simplex method), one-way and two-way ANOVA, Principal Component Regression.

  • Statistical charts

    Optional with registered TeeChart/TeeChart.NET: Probabilities plot (Normal, Weibull, QQ), variable control charts ( X, R, S and EWMA), attribute control charts (P, NP, U and C), dot plot, box plot, biplot, error ellipses PLUS all major statistical chart types, supported by Steema's TeeChart (error, barr, error bar, pie, box, scatter, scatter 3D, histogram, Pareto and more).

  • Time series analysis

    Sample ACF, PACF, exponential smoothing (single, double, triple), support for ARMA/ARIMA models (simulating, forecasting, estimating coefficients by using Yule-Walker, Burg, Innovations and MLE algorithms), ARAR time series model, moving average, memory-shortening filter, Box-Ljung statistics, etc..

  • Ready-to-use components

    TMtxANOVA (encapsulates ANOVA routines), TMtxMulLinReg (encapsulates multiple linear regression routines), TMtxNonLinReg (encapsulates nonlinear regression routines), TMtxPCA (encapsulates PCA routines), TMtxHypothesisTest (encapsulates hypothesis testing routines), TMtxBinaryTest (encapsulates binary diagonstic test routines), TMtxMDScaling (encapsulates multimensional scaling routines).

For more information on Stats Master, the statistical add-on package for MtxVec math library:

Continue reading