Statistics

Note: sv_simfit has a simplified version of this program with fewer options

Abstract

The main Simfit statistical program is called Simstat and there are test files to demonstrate every option. Also, note that the reference manual contains the theory for all of the procedures supported by Simfit together with worked examples. In addition there are numerous dedicated programs that often have additional statistical options.

Program Simstat

  1. The Simstat data exploration options
    Exhaustive-analysis: arbitrary vector
    Exhaustive-analysis: arbitrary matrix
    Exhaustive-analysis: multivariate-normal matrix
    All possible pairwise MWU/KS2/t-tests
    Robust analysis: one sample
    Robust analysis: two samples
  2. The Simstat standard tests options
    1-sample t test
    1-sample Kolmogorov-Smirnov test
    1-sample normal distribution test
    1-sample Poisson distribution test
    2-sample unpaired t test
    2-sample paired t test
    2-sample Kolmogorov-Smirnov test
    2-sample Wilcoxon-Mann-Whitney-U test
    2-sample Wilcoxon signed-rank test
    Chi-sq./Fisher-exact/log-linear contingency table
    McNemar test on paired frequencies
    Cochran Q test (on 0/1 integer matrix)
    Binomial test: K successes in N trials
    Sign test: known number of +/- signs
    Runs test: known number of +/-, runs
    F test for excess variance: two WSSQ
  3. The Simstat Analysis of Variance options (ANOVA)
    Bartlett and Levene tests for homogeneity of variance
    1-way and Kruskal-Wallis nonparametric
    1-way only
    1-way Kruskal-Wallis only
    2-way and Friedman nonparametric
    2-way only
    2-way Friedman only
    Latin Square
    Groups and subgroups
    Factorial design
    Repeated measures
  4. The Simstat Analysis of Proportions options
    Analysis of proportions
    Cochran-Mantel-Haenszel Meta Analysis
    Bioassay, Dose response and LD50
  5. The Simstat multivariate statistics options
    Correlation: Pearson Product Moment
    Correlation: Spearman and Kendall-tau
    Correlation: canonical (2 subgroups)
    Correlation: partial (>2 variables))
    Clusters: dendrograms (arbitrary matrix)
    Clusters: scaling from (arbitrary matrix)
    Clusters: dendrograms and scaling (distance matrix)
    Clusters: K-means
    Principal components analysis: PCA
    Rotation: Procrustes analysis
    Rotation: Varimax or Quartimax
    Compare groups: MANOVA/means/profiles
    Compare groups: canonical variates/PCA
    Compare groups: distances/allocations
    Factor analysis
    Biplots
  6. The Simstat regression options
    Fit a line (simple least squares)
    Fit a line (simple reduced major axis)
    Fit a line (simple orthogonal)
    Fit a line (advanced least squares)
    Fit a line (advanced reduced major axis)
    Fit a line (advanced orthogonal)
    Fit a line/calibrate (simple)
    Fit a line/calibrate (advanced)
    Fit a polynomial/calibrate (x,y)
    Fit a polynomial/calibrate (g(x),f(y))
    Fit a multilinear model f(x1,x2,...,xn)
    Fit a dose-response curve (LD50 by GLM)
    Fit logistic regression models (by GLM)
    Fit Cox proportional hazards model
    Compare 2 regression parameters
    Compare 2 sets of regression parameters
  7. The Simstat Generalized Linear Model Options
    Comprehensive GLM options
    Logistic regression
    Binary logistic regression (no strata)
    Binary logistic regression (with strata)
    Polynomial logistic regression
    Exponential survival
    Weibull survival
    Extreme value survival
    Cox proportional hazard survival
    Contingency table analysis
    Bioassay (percentiles/EC50/LD50)
  8. The Simstat time series and survival options
    Data smoothing
    Lags, ACF and PACF
    Fit and predict by ARIMA
    Survival curves (Kaplan-Meier)
    Survival analysis (GLM techniques)
    Cox proportional hazard analysis
  9. The Simstat statistical calculations options
    Statistical power and sample size
    Estimate parameter confidence limits
    Robust calculations for one sample
    Robust calculations for two samples
    Shannon/Brillouin diversity indices
    Plot/evaluate Standard distributions
    Plot/evaluate Non-central distributions
    Quantify ligand binding cooperativity
    Random numbers/permutions/Latin-Squares
    Kernel density estimation
  10. The Simstat numerical analysis options
    Polynomial: calculate zeros
    Matrix: determinant/eigenvalues/inverse
    Matrix: singular value decomposition
    Matrix: LU factorisation/norms/cond.no.
    Matrix: QR factorisation
    Matrix: Cholesky factorisation
    Solve: Ax = b (A nonsingular)
    Solve: Ax = b (L_1 norm overdetermined)
    Solve: Ax = b (L_2 norm overdetermined)
    Solve: Ax = b (L_i norm overdetermined)
    Calculate: (y^T)Ay, (y^T)(A^{-1})y
    Calculate: AB,(A^T)B,A(B^T),(A^T)(B^T)
    Calculate: Ax = lambda*Bx (B pos.def.)
    Rotation: Orthomax
    Rotation: Procrustes

Advanced statistics programs with additional options

Back to Help Menu or End Help