Gretl is a free and open source package for econometric analysis on 15-04-2017 updated version gretl2017a was released. The following are the new features, bug fixes and change log of new version.

  • Improved parsing for the member-of relation, on both the left- and right-hand side of hansl statements
  • Calendar: ensure that the Gregorian calendar is used consistently but add some optional support for handling Julian dates (also add a User’s Guide chapter on this)
  • xls, xlsx, ods and gnumeric data importers: handle string data (this facility extended from the CSV importer)
  • Remove obsolete Kalman-filter interface
  • Sequential numbering of models: skip models estimated with the –quiet flag and also those estimated within functions
  • Deprecate “=” as obsolete synonym for boolean “==”
  • New functions kmeier() and naalen() for nonparametric estimation of survivor and hazard functions
  • Add normtest() function: works much like the “normtest” command, but handles vectors as well as series
  • Add ecdf() function to compute the empirical CDF for a
      series or vector
  • New function npcorr() for nonparametric correlation, either Kendall’s tau or Spearman’s rho
  • bwrite() function for writing bundles as XML: support gzip compression if the filename has the “.gz” extension
  • typeof() function: document it, and extend it to cover members of bundles and arrays
  • “include” command: make it respect the “catch” flag, and add a –force option to force re-reading of gfn files
  • “corr” command: add a –plot=filename option; also add matrix option to take data from a named matrix
  • “gnuplot” and “plot” commands: add a –font option
  • “fcast” command: in the case of the –rolling option, accept the name of a scalar variable for the steps-ahead argument; also respect the last observation specification in all cases
  • “tsls” command: calculate Sargan test in overidentified case even when there are no endogenous regressors, to allow “difference in Sargan” test
  • “modtest” command with –autocorr or –arch option, as applied to VAR/VECM: switch from per-equation tests to multivariate tests as described in Lutkepohl’s 2005 book, but add a –univariate option to give per-equation tests
  • “var” and “vecm”: enable the –window (-w) option to show results in a window from script execution
  • “hausman” command: add options –nerlove and –matrix-diff to inflect the random effects estimator and calculation of the Hausman test statistic
  • Principal Components window in GUI: support selection of a specific number of components to save as series
  • ARMA models in GUI: add menu item to view plot of spectrum versus sample periodogram
  • Bug fixes: functions failing to return loop-index value, couldn’t assign to an array member of a bundle, multiple X-Y scatterplots could fail when using long variable names, the matrix-oriented version of the “restrict” command would fail when only the constant was restricted,reading of gzipped matrix files via mread() was much slower than necessary for large matrices, couldn’t read .mat (matrix) files generated by R on Windows when they contained NaN values, closing the function package browser while a package is running could cause a crash, possibly misaligned labels when doing a group- means scatterplot using panel data, “outfile” command with –buffer option was failing on MS Windows
  • Update documentation for inlist() function
  • Update translated Command Reference PDF files for Italian and Portuguese
  • Add Galician (Galego) translation of the online Function Reference
  • User’s Guide: add a chapter on degrees of freedom correction for standard errors and related statistics
  • Build for Mac OS X: update the bundled GTK stack
  • Builds for MS Windows: update compiler to gcc 5.4.0

Subscribe to Sulthan Academy, stay connected and receive updates.

Leave a comment

Your email address will not be published. Required fields are marked *