What's new in DataLab 4.000
Feb 8, 2021
- New Features:
- DLabPascal now provides an integrated debugger
- several extensions of the built-in DLabPascal library
- signal integration tool
- a tool for calculating binary classifier metrics
- new statistical test: median test
- Improvements and Changes:
- Wilcoxon test report now displays the median of the differences as well
- redesigned random forest form
- a variable set repository makes the selection of variables much more convenient
- the RF applicator has now an option to create binary results for classifiers
- the PLS Applicator now shows the CF thresholds and allows to store the results in the aux. maxtrix or append it to the data matrix
- the data editor now supports ^C and ^V for copy and paste (instead of 'extract from clipboard')
- refurbished k-means clustering
- RBF network extended
- PLS/DA now displays more classifier metrics (same as MBL)
- numerical editor: row names are now displayed using a much longer maximal length; two additional fonts (Tahoma, Verdana) included
- MLR and Ridge Regression diagrams provide now more plot options
- memory based learner improved in many details
- refurbished Kohonen map
- improved edit/selection mode in data description window
- additional performance metrics for RF classifiers
- formula editor now uses the ° symbol as cell reference
- class color setup dialog now supports saving and loading color palettes
- the default class color palette has been improved
New in DataLab 3.530 (Nov 28, 2013)
- New Features:
- ridge regression implemented
- import/export of data in XML fomat
- Improvements and Changes:
- update server reconfigured (user have to install from scratch)
- new option to prevent the multiple drawing of symbols on top of each other
- a few minor cosmetic improvements
- MLR model offers now more options to store the results
- the variable selection dialog allows now multiple selections
- script engine extended to support ridge regression
- the calculation of correlation functions can now be aborted
- box plots may be displayed now directly from the main menu
- the statistical tests now support addtional commands to exchange red and blue markings
- Bug Fixes:
- cross validation of PLS models with more than 15 target variables resulted in range error
- correlation function: reentrancy problems fixed
- Spearman's rank correlation delivered wrong results for large data sets (>50000 values)
- univariate descriptors showed floating point error on empty data matrix
- F-Test: incorrect degrees of freedom were calculated if the variance of the blue group was larger than the variance of the red group
- reg. license was not recognized correctly in some special cases
- "Revert" command did not revert the comment displayed in the matrix viewer
New in DataLab 3.511 (Mar 16, 2013)
- the link into the help page of the 2D data designer was invalid
- the symbol table did no show any symbols
New in DataLab 3.510 (Mar 15, 2013)
- New Features:
- 2D dataset designer
- the data filtering now includes a moving median filter
- range of axes can now be copied from one plot to another by drag&drop
- Improvements and Changes:
- displaying speed of charts with a huge number of data points (as symbols) is increased by a factor of 10
- import of CSV data provides now a progress feedback
- speed of loading data files improved
- a warning is displayed if the user opens an IDT file which is newer than the current version of DataLab
- serialisation of data matrices now includes the column and row index of the original data
- 3D surface plot offers now two types of interpolation
- MLR of large data sets is now much faster
- Bug fixes:
- various minor bug fixes
- plot window displayed error message if one of the variables of a bivariate plot was empty
- new windows are now positioned correctly in a multi-monitor environment
- text import filter: overreading intial lines now works even if trigger is empty
New in DataLab 3.507 (Jan 15, 2013)
- Improvements and Changes:
- PCR window provides now a context menu for the various charts
- speed of cross correlation function increased
- the calculation of filtered data can now be restricted to a particular range of the data matrix
- Bug Fixes:
- rowwise filtering of data did not write the results back to source matrix
- various minor bug fixes
New in DataLab 3.506 (Jan 7, 2013)
- New Features:
- DataLab provides now a simple way to extract data from text files
- Pareto Charts
- Improvements and Changes:
- the dialog for selecting the working directory has been improved
- double-clicking the data plot switches to a list of the most important statistical parameters
- XLS import routine now recognizes categorial columns automatically and allows to specify date/time format for headings
- text file extraction improved (mostly cosmetic improvements)
- image digitization provides now a button to paste an image from the Windows clipboard
- results of the filter process can now be stored in the matrix clipboard, as well
- data imputation can now be applied row-wise, as well
- plot window now offers an option to display date/time coordinates
- DataLab now supports the direct import of both Excel XLS and XLSX files
- MS Sans Serif is replaced by Arial font throughout the program to get more space for labels etc.
- data editor now offers three additional button to scroll the spreadsheet
- various cosmetic changes
- the matrix clipboard can now hold up to nine auxiliary matrices (copy of the original + 8 user defined)
- the original data matrix is now automatically stored in matrix clipboard upon load from disk
- the work matrix can now be selected from the matrix clipboard
- data editor now displays the range and number of selected cells
- random number generator: user defined profiles can now be saved to disk
- PLS-DA: classification results now display various performance indicators
- PLS-DA now provides cross validated true-positive- and false-positive rates
- DataLab blog is now integrated into support menu
- heading generator now generates month and day names using the DataLab language setting (and not the locale)
- line and spectral plots now provide the option to indicate the data points
- the dendrogram can now be displayed as numerical table and in Newick format
- the help file now offers links to the corresponding pages of the DataLab-Blog (German only)
- many small, mostly cosmetic changes
- filter command can now be applied to several columns or rows and offers additional features (smoothed second derivative acc. to Savitzky-Golay, selection of storage mode)
- script interpreter supports two additional commands ('filtercoeffs' and 'filter')
- the speed of PLS discriminant analysis with very large data matrices has been increased by 20 to 50%
- both the variable and the object selector support now the selection of data at regular intervals
- PLS indicates the progress of a time-consuming calculation in a progress bar
- help improved
- Bug Fixes:
- a missing working directory does not result in a program crash any more
- cluster analysis crashed if the matrix size has been changed while dendrogram window was open
- the distances of Ward's method of cluster analysis are now calculated correctly
- Durbin-Watson test in univariate Regression is now correctly applied (problem with unsorted data)
- percentiles for vary small samples are now calculated correctly
- empty cells were set non-empty when copying from the matrix clipboard to the main matrix
- dendrogram now displays class colors after opening the cluster window
- communication between data editor and data imputation window did work as expected
- the final columns were not processed correctly when filtering the data rowwise
- the level of measurement has been processed incorrectly when deleting marked columns
- memory overflow fixed (occurred when displaying boxplots of marked data of huge matrices)
- Durbin-Watson test in univariate Regression is now correctly applied (problem with unsorted data)
- script commands "markdata" and "unmarkdata" did not process parameter list correctly
- univariate regression: object indicator now displays correct row name for marked data, as well
- nominal identifiers are now correctly reset when reading IDT files
- aborting the update process now provides appropriate feedback
- marking the data in the test window did not affect the main data matrix
- refresh of class symbols did not work when assigning different class symbols or class numbers
- a few minor cosmetic bug fixes
- import from XLS file: comment is now correctly initialized
- the logarithmic display of the FFT plot generated a floating point error
- the background of main window is now drawn correctly even if DataLab spans several screens
- invoking the correlation function or a statistical test via a script now correctly translates these windows
- several translation errors in the German version fixed
- the update dialog created a weird error message if a proxy between the user system and the update server required an authentication
- changing the option "Use full precision when copying to clipboard" did not have any effect as long as the numeric data editor was open
- control to set the filter size did now work properly
- missing refresh message to the other windows after executing a script
- matrix viewer did not refresh its image when empty state of cells changed
- histogram of a column/row displayed a prompt to mark the data if the corresponding column/row was empty
- normal probability plot raised an exception if the number of select data was too low
New in DataLab 3.505 (Dec 18, 2012)
- Improvements and Changes:
- double-clicking the data plot switches to a list of the most important statistical parameters
- XLS import routine now recognizes categorial columns automatically and allows to specify date/time format for headings
- text file extraction improved (mostly cosmetic improvements)
- image digitization provides now a button to paste an image from the Windows clipboard
- results of the filter process can now be stored in the matrix clipboard, as well
- Bug Fixes:
- Durbin-Watson test in univariate Regression is now correctly applied (problem with unsorted data)
- percentiles for vary small samples are now calculated correctly
- empty cells were set non-empty when copying from the matrix clipboard to the main matrix
New in DataLab 3.504 (Dec 18, 2012)
- New Features:
- DataLab provides now a simple way to extract data from text files
- Improvements and Changes:
- data imputation can now be applied row-wise, as well
- plot window now offers an option to display date/time coordinates
- DataLab now supports the direct import of both Excel XLS and XLSX files
- MS Sans Serif is replaced by Arial font throughout the program to get more space for labels etc.
- data editor now offers three additional button to scroll the spreadsheet
- various cosmetic changes
- Bug Fixes:
- dendrogram now displays class colors after opening the cluster window
- communication between data editor and data imputation window did work as expected
- the final columns were not processed correctly when filtering the data rowwise
- the level of measurement has been processed incorrectly when deleting marked columns
- memory overflow fixed (occurred when displaying boxplots of marked data of huge matrices)
- Durbin-Watson test in univariate Regression is now correctly applied (problem with unsorted data)
New in DataLab 3.503 (Dec 18, 2012)
- New Features:
- Pareto Charts
- Improvements and Changes:
- the matrix clipboard can now hold up to nine auxiliary matrices (copy of the original + 8 user defined)
- the original data matrix is now automatically stored in matrix clipboard upon load from disk
- the work matrix can now be selected from the matrix clipboard
- data editor now displays the range and number of selected cells
- random number generator: user defined profiles can now be saved to disk
- PLS-DA: classification results now display various performance indicators
- PLS-DA now provides cross validated true-positive- and false-positive rates
- DataLab blog is now integrated into support menu
- heading generator now generates month and day names using the DataLab language setting (and not the locale)
- line and spectral plots now provide the option to indicate the data points
- Bug Fixes:
- script commands "markdata" and "unmarkdata" did not process parameter list correctly
- univariate regression: object indicator now displays correct row name for marked data, as well
- nominal identifiers are now correctly reset when reading IDT files
- aborting the update process now provides appropriate feedback
- marking the data in the test window did not affect the main data matrix
- refresh of class symbols did not work when assigning different class symbols or class numbers
- a few minor cosmetic bug fixes
New in DataLab 3.502 (Aug 28, 2012)
- Improvements and Changes:
- the dendrogram can now be displayed as numerical table and in Newick format
- the help file now offers links to the corresponding pages of the DataLab-Blog (German only)
- many small, mostly cosmetic changes
- Bug Fixes:
- import from XLS file: comment is now correctly initialized
- the logarithmic display of the FFT plot generated a floating point error
New in DataLab 3.501 (Aug 21, 2012)
- Improvements and Changes:
- filter command can now be applied to several columns or rows and offers additional features (smoothed second derivative acc. to Savitzky-Golay, selection of storage mode)
- script interpreter supports two additional commands ('filtercoeffs' and 'filter')
- the speed of PLS discriminant analysis with very large data matrices has been increased by 20 to 50%
- both the variable and the object selector support now the selection of data at regular intervals
- PLS indicates the progress of a time-consuming calculation in a progress bar
- help improved
- Bug Fixes:
- the background of main window is now drawn correctly even if DataLab spans several screens
- invoking the correlation function or a statistical test via a script now correctly translates these windows
- several translation errors in the German version fixed
- the update dialog created a weird error message if a proxy between the user system and the update server required an authentication
- changing the option "Use full precision when copying to clipboard" did not have any effect as long as the numeric data editor was open
- control to set the filter size did now work properly
- missing refresh message to the other windows after executing a script
- matrix viewer did not refresh its image when empty state of cells changed
- histogram of a column/row displayed a prompt to mark the data if the corresponding column/row was empty
- normal probability plot raised an exception if the number of select data was too low
New in DataLab 3.500 (Aug 21, 2012)
- New Features:
- update window now provides a link to a Web page showing the latest changes
- PLS-discriminant analysis implemented
- dichotomisation tool implemented
- new command "Data/Edit/Set to Zero/...." allows to set parts of the data matrix to zero values
- new command "Create Indicator Variables" supports the generation of binary indicator variables
- the new command "File/Recently Opened" allows to select from a list of previously opened files
- Improvements and Changes:
- the scatter plot window now displays marked data and allows to mark data of both types
- process communication between various windows improved
- help file improved
- markings can now be exchanged using the numerical table editor
- handling of logarithmic scale labels improved
- colors of marked data are now adjustable by the user
- histogram of nominal/ordinal variables now create a point histogram (one variable per bin) if class definition is set to automatic
- ordinal and nominal scales now show the ordinals in parantheses if no identifiers are defined
- PCR now has a user interface which resembles the MLR interface and allows to mark data
- IDT format extended to include range mode of histograms, cumulative frequency plots and normal prob. plots
- cumulative frequency plots for small data collections are now based on sorted data (and not on the histogram bins)
- histogram setup now offers an automatic selection of the number of bins
- cross correlation table now allows to switch between different measures of correlation
- eyedroppers now invoke variable/object selection dialog; selection via matrix survey plot is no longer available
- digitisation supports now logarithmic scales as well
- protocols for the standards tests now display probabilities, as well
- if two blocks are paired, the tests for normality now also test the differences of two blocks
- file open/save dialogs now offer a filter option to narrow down the list of available files
- "Split Data" command now allows to create test and training data sets and offers file template
- PCA - speed of display of scores is now improved for large matrices
- data editor now supports the inversion of markings
- PLS cross validation protocol now includes the mean error traces of the individual target variables
- the range of target values when applying a PLS model is now checked for a possible overlap with the descriptors
- comments of PLSM files are now displayed in the file dialog
- processing and user interface of correlation function improved
- ANOVA now displays p-values and allows to create a protocol
- deletion of files now provides feedback
- cases of object and variable names are now ignored during sort
- various minor improvements
- PCA and PCR now take care of the scaling mode when indicating important components
- misleading option "adjust matrix size after delete" discarded
- initial startup process improved
- Bug Fixes:
- some minor bug fixes
- colors in color selection boxes show now readable names (instead of technical colors constant identifiers)
- DataLab now checks for file versions which are newer than the current program version
- histogram/cumulative frequ./normal prob. plots of marked data now work as expected if no data are marked
- cross correlation table does no longer indicate wrong minimum/maximum values when switching from one to another table
- when loading images for digitisation via Windows explorer the BMP-files are now displayed in the open dialog
- cross validation in PLS models did not select a random test set (but a consecutive block of data)
- univariate statistics: all variables can now be selected (was limited to the first 20 variables)
- levels of measurements have not been copied to and from auxiliary matrices
- forbidden access to update server via proxy or gateway which requires login is now correctly recognized
- Durbin-Watson test of regression residuals is now correctly classified if DW statistic is above 2.0
- marking of data retriggered initialisation of PCA window
- fluent editing of names and class numbers has almost been blocked by a faulty broadcast event
- MLR variable selection window did not react on changes of the data matrix layout
- when storing a PLS model, the names of the variables have been lost under certain circumstances
- calculating univariate properties resulted in a runtime error when the matrix has less than 6 columns
- PCR checks now for validity of variables
- program crashed if size of data matrix was set to a length/width of greater than 2^31
- program crashed if entire matrix was deleted
- improved validity check of Windows registry avoids crashes in the base of corrupt registry entries