Programje

Statistical software developed at the Institute of Biomedical Informatics, Faculty of Medicine, University of Ljubljana:

• re R function for calculating the Re measure of explained variation for the Cox model, Aalen's model or any of the parametric models included in the survreg function
  • function
  • programmed by Maja Pohar Perme
• pseucheck R function for model diagnostics using pseudo-observations: for models fitted with 'coxph', 'crr', 'lm' and 'geese' function
  • function
  • requires packages: pseudo, survival, cmprsk, geese
  • programmed by Maja Pohar Perme

• koq function for S/S-PLUS (available at StatLib)
  • calculates Kent & O'Quigley's measure of dependence for censored data when dependence between time and covariates is modelled by Cox's model
  • programmed by Andrej Blejec & Janez Stare
• bj function for S/S-PLUS/R (incorporated into Design package by Frank E. Harrell)
  • fits the Buckley-James distribution-free least squares multiple regression model to a possibly right-censored response variable
  • programmed by Janez Stare, Harald Heinzl & Frank E. Harrell
• measures of explained variation in survival analysis (other than koq) for S/S-PLUS (available upon request)
• pseu R function for Cox and additive model diagnostics using pseudo-observations for producing plots described in Pohar Perme M, Andersen PK. Checking hazard regression models using pseudo-observations. Statistics in Medicine 2008;27:5309-5328.
  • function
  • help file
  • requires the following libraries: survival, Design, geepack (all available from CRAN) and timereg (available from http://staff.pubhealth.ku.dk/~ts/timereg.html)
  • programmed by Maja Pohar
• relsurv package for relative survival analysis in R (version 2.0, included as contributed package into CRAN)
  • implementation of existing regression methods plus our own method (Stare, Henderson & Pohar: An individual measure of relative survival. JRSS Series C, 2005, 54(1), 115-126)
  • implementation of goodness of fit methods based on residuals (Stare, Pohar & Henderson: Goodness of fit of relative survival. Statistics in Medicine, 2005, 24(24), 3911-3925)
  • paper describing the package (Pohar & Stare: Relative survival analysis in R. Computer Methods and Programs in Biomedicine, 2006, 81(3), 272-278)
  • programmed by Maja Pohar
• interactive demonstration of relative survival
  • website (form with links to instructions and references)
  • based on the individual measure of relative survival and the relsurv package (see above)
  • idea by Janez Stare, implementation by Brane Leskošek (Perl, Linux) and Maja Pohar (R)
• R functions for calculating the reduced citation distribution (h-index) function
  • function
  • programmed by Maja Pohar Perme
• chplot package for R (current version 1.2, included as contributed package into CRAN)
  • produces augmented convex hull plots, which are a nice and informative way of displaying large amounts of grouped bivariate data
  • paper describing the package (Vidmar & Pohar: Augmented convex hull plots: Rationale, implementation in R and biomedical applications. Computer Methods and Programs in Biomedicine, 2005, 78(1), 69-74)
  • package source (tar.gz)
  • binary package for Windows (zip)
  • idea and design by Gaj Vidmar, programmed by Maja Pohar
• visualisation of concordance
  • two newly proposed graphical displays for the previously unaddressed task of visualising concordance (for ranked data, as for, e.g., Kendall's W)
    • concordance bubble-plot (Excel workbook with macro and sample data) -- depicts raw data, i.e., the actual ranks assigned to objects
    • pin-cushion plot (interactive and batch-mode code with sample data for jsplot; zipped) -- depicts within-object rank differences
  • paper describing the methods (Vidmar & Rode: Visualising concordance. Computational Statistics, 2007, 22(3), in press)
  • concordance parallel-coordinates-plot (SigmaPlot 2000 notebook; zipped) -- another possibility (depicts pairs of ranks assigned to objects), which proved to be less practically useful (though esthetically appealing)
  • ideas, design and implementation by Gaj Vidmar & Nino Rode
• Bland-Altman function for R
  • displays the classic Bland-Altman difference plot with limits of agreement, prints the mean difference and both agreement limits (along with 95% CIs) and returns a list of relevant parameters (the mean difference, its SD and SE, upper and lower agreement limits, their SE, and the t value used in the calculation of CIs)
  • function and sample data (zip)
  • programmed by Jaro Lajovic
• Excel workbooks for demonstrational and teaching purposes
  • Poisson regression
    • estimation procedure demonstrated with example 4.5 from A.J.Dobson, An Introduction to Generalized Linear Models (1st ed.), Chapman & Hall, 1990
    • based on Excel's matrix functions, step-by-step iteration with macro
    • requires Excel 97 or higher
    • idea and implementation by Gaj Vidmar
  • parallel coordinates plot
    • static version for producing presentation/publication graphics
    • for two variables, typical use is in conjunction with paired-samples t-test, but such plot can also accompany the Bland-Altman method (see below)
    • requires Excel 97 or higher
    • idea and implementation by Gaj Vidmar
  • modified Box-Cox power transformation towards normal distribution
    • the lambda parameter can be either input manually or estimated with the maximum-likelihood method using Excel's Solver add-in
    • the transformation effect is demonstrated with automated histogram
    • requires Excel 97 or higher
    • idea and implementation by Gaj Vidmar
  • interactive population pyramid (Eurostat's projection for Slovenia 2004-2051)
    • the words are in Slovenian, but they are very few and with the image, everything should be understandable for anyone speaking any European language
    • data obtained from the SI-STAT database of the Statistical Office of the Republich of SLovenia
    • inspired by the demographical model movies by Erich Neuwirth (University of Vienna, Austria)
    • requires Excel 97 or higher
    • implementation by Gaj Vidmar
  • Bland-Altman plot
    • implements the classic, simple, indispensable yet still too often overlooked method-comparison procedure by Bland & Altman
    • requires Excel 97 or higher
    • core code by Michael Schacht Hansen (Section for Health Informatics, University of Aarhus, Dennmark); workbook, error trapping and embellishment by Jaro Lajovic & Gaj Vidmar
  • windowgram and coplot
    • presents two methods, endorsed by Larry Weldon (Simon Fraser University, Vancouver, Canada) as under-used but simple and ideal for inclusion into introductory statistics courses
    • windowgrams are the simplest case of and hence the introduction to kernel density estimation; coplots is short for conditional plots, so it is just another name for what is also called panel plots, Trellis display or lattice graphics
    • requires Excel 97 or higher
    • idea and implementation by Gaj Vidmar
  • dot plots
    • presents simple and multi-way dotplots as invented by William S. Cleveland and promoted by S-Plus and R
    • requires Excel 97 or higher
    • idea and implementation by Gaj Vidmar
  • univariate dot-density plot
    • presents the simplest yet very nice solution, i.e, character-based cell-chart, of what is also known as stacked dot-plot
    • requires Excel 97 or higher
    • idea and implementation by Gaj Vidmar
  • stem-and-leaf plot
    • requires Excel 97 or higher
    • macro by Nick Maxwell (Data Matters Resource Center); debugging and workbook design by Gaj Vidmar
  • Kaplan-Maier survival curve estimation
    • requires Excel 97 or higher
    • unauthorised English translation and modification of the workbook with macro by Shigenobu Aoki (Faculty of Social and Information Studies, Gunma University, Japan); produced by Gaj Vidmar
  • automated histogram binning (with random sampling from normal distribution)
    • requires Excel 97 or higher
    • idea and implementation by Gaj Vidmar
  • Iris dataset
    • the most (perhaps too) often used (and sometimes abused) dataset in the field of statistics (too often abused in the fields of machine learning, data mining and information visualisation)
    • reference for data collection: Anderson, E. (1935). The irises of the Gaspe Peninsula. Bulletin of the American Iris Society, 59, 2-5.
    • reference for data analysis: Fisher, R.A. (1936). The use of multiple measurements in taxonomic problems. Annals of Eugenics, 7 (2), 179-188.
    • a truly comprehensive presentation: Les Iris de Fisher ou Comment se familiariser avec le logiciel R (in French)

Ostalo programje ki smo ga razvili:

• PowerPointov makro za štetje vseh znakov v predstavitvi
  • sprogramiral Gaj Vidmar
  • razvit za potrebe prevajalcev, da olajša zaračunavanje prevodov predstavitev
  • poleg znakov v diapozitivih, zna makro prešteti tudi znake v matricah diapozitiva, opomb, izročkov in naslova
  • deluje z angleško in slovensko verzijo Microsoftove pisarne
  • makro je namenjen slovenskim uporabnikom, zato je uporabniški vmesnik v slovenščini in tudi navodila za namestitev/odstranitev (spodaj) so le v slovenščini
  • programska koda je zaščitena z geslom
  • namestitev in odstranitev:
    • datoteko StetjeZnakov.ppt shranite na disk
    • zaženite PowerPoint in datoteko odprite, nato pa jo shranite kot dodatek (.ppa)
    • kam se bo dodatek shranil, je odvisno od konfiguracije računalnika (npr. na Windows® 2000 v mapo C:\Documents and Settings\uporabniško_ime\Application Data\Microsoft\Addins)
    • zaprite PowerPoint
    • ponovno zaženite Powerpoint ter poskrbite, da so makri omogočeni (Orodja - Možnosti - Varnost - Varnost makrov... - nizka ali srednja)
    • namestite nov dodatek (Orodja - Dodatki... - Dodaj nov...)
    • v meniju Orodja se bo pojavila opcija "Štetje znakov"
    • če ga ne želite več uporabljati, dodatek preko menija, iz katerega ste ga namestili, odstranite iz pomnilnika ali povsem odstranite
    • tudi, če izberete popolno odstranitev, bo datoteka StetjeStrani.ppa ostala na disku, zato jo morate, če to želite, izbrisati ročno iz zgoraj omenjene mape