Home | People | Partners | VOStat | StatCodes | Data & Tutorials | Programs | Bibliographies

VOStat (rev 2) Statistical Analysis for the Virtual Observatory Start using VOStat right now

VOStat: FAQ

Transforming and subsetting the data

• Select cases
• Simple transform

Plots and summaries

• Boxplot: A univariate plot giving a rough idea of the distributions and outliers.
• Dotplot: Plotting values of the same variable for different objects (e.g., densities of different asteroids).
• Histogram: Univariate histogram.
• Averaged Shifted Histogram (ASH): Univariate ASH.
• Scatterplot: A bivariate plot allowing error bars.
• Make a pairs plot. This is sometimes also called a ScatterPLOt Matrix or SPLOM.
• 3D scatterplot
• Summary statistics

Estimating distribution from data

• Histogram
• Averaged Shifted Histogram (ASH): Univariate ASH.
• Empirical distributon function (ECDF)
• Kernel density estimation:
  • 1-dimensional (fixed window width)
  • 1-dimensional (adaptive window width)
  • 2-dimensional
• Fiting parametric distribution

Comparing data with distributions

• Normal plot
• Tests for normality
• 1-sample KS test. This is useful for checking if a variable in your data set is distributed according to some standard distribution.
• Anderson-Darling test. This is another test for checking if a variable in your data set is distributed according to some standard distribution.
• 2-sample KS test to compare the distributions of two continuous variables in a data set.
• Chi-square 2 sample test to compare the distributions of two discrete variables in a data set.

Some standard tests

• One sample z-test to compare the mean of a variable with some given number. This test is applicable only if you know (or have a very good approximation for) the standard deviation of your variable. (Requires Gaussianity assumption.)
• One sample t-test. Similar to above, except that you do not need to know the standard deviation. (Requires Gaussianity assumption.)
• paired t-test for comparing the means of two variables. (Requires Gaussianity assumption.)
• Wilcoxon's signed rank test (1 sample). Much like one sample t-test, but without the Gaussianity assumption.
• Wilcoxon's signed rank test (2 sample). Much like paired t-test, but without the Gaussianity assumption.

Regression

• Linear regression
• Nadaraya Watson local regression
• Major axis/orthogonal regression
• LOESS
• Linear quantile regression
• Robust regression using M estimation
• Robust regression using trimmed least squares
• Theil-Sen regression

Multivariate methods

• Make a pairs plot. This is sometimes also called a ScatterPLOt Matrix or SPLOM.
• Linear regression
• Principal component analysis
• Canonical correlation analysis
• Model based clustering
• Hierarchical clustering

Spatial methods

• Variogram
• k-nearest neighbours and Moran's I testa
• Ripley's K and related measures
• Voronoi tesselation and adaptive 2D kernel density estimation

Directional statistics

• Plotting directional data
• Summarizing direction data
• Kernel density estimation
• Fitting a von Mises distribution
• Testing goodness-of-fit with a von Mises distribution
• Testing goodness-of-fit with uniform distribution

Survival analysis

• Maximum likelihood Kaplan-Meier estimate of the survival curve for censored data
• A bunch of tests for censored data: Log rank, Peto-Peto, Kolmogorov-Smirnov, Cramer-von Mises, and Anderson-Darling tests
• Maximum likelihood Lynden-Bell-Woodroofe estimator

Time series analysis

• Time series plotting with smoothing
• (Partial) AutoCorrelation Function (ACF and PACF)
• Autoregressive modelling
• Periodogram with smoothing option

VOStat uses different output formats for different purposes:

• graphical output: png, eps and pdf.
• tabular output: VO table (downloadable by [right/command]-clicking or via SAMP)
• R scripts: ASCII text, that you can copy-n-paste from the webpage.
• R workspace: Binary R format.

samp.app.ping
samp.hub.disconnnect
samp.hub.event.metadata
samp.hub.event.shutdown
samp.hub.event.subscriptions
samp.hub.event.unregister
client.env.get
table.load.votable