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
- 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
- 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
- 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
- The Simstat Analysis of Proportions options
Analysis of proportions
Cochran-Mantel-Haenszel Meta Analysis
Bioassay, Dose response and LD50
- 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
- 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
- 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)
- 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
- 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
- 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
- Program RSTEST
Nonparametric statistics, e.g., runs, signs, Kolmogorov-Smirnov,
Mann-Whitney, Kruskal-Wallis, Friedman, regression on ranks, etc.
- Program BINOMIAL
The binomial, Poisson, and non-central beta distributions.
- Program CHISQD
The chi-square and non-central chi-square distributions.
- Program FTEST
The F and non-central F distributions.
- Program RANNUM
Pseudo random vectors, matrices, Latin squares, and random
walks in 1, 2, or 3 dimensions.
- Program LINFIT
Comprehensive linear regression techniques, e.g., L_1, L_2,
L_infinity, Robust (M-estimates).
- Program NORMAL
The normal distribution.
- Program GCFIT
Nonlinear growth curves, survival curves, and survival analysis
- Program TTEST
The t and non-central t distributions.
- Program SPLINE
Fitting least squares, smoothing, or cross-validatory splines
to calibrate, or estimate derivatives, areas, arc length,
curvatures, etc.
Back to Help Menu or End Help