Ginkgo Log
Version 1.5.8 - (18/jun/2008) Latest version
Evaluating remaining objects in distance-based discriminant analysis
Posidonia oceanica multivariate index (POMI) implemented
Implementation of Caillez correction of PCoA negative eigenvalues
Implementation of Fuzzy Rand Index as published in Campello(2007)
Version 1.5.5 - (07/jun/2007)
Improving JCOMPAS menus & operability
NMDS is now MDS with several options
Non-metric (ordinal) MDS available for similarities and distances
Ratio MDS available for distances
Local MDS options for focusing on concrete objects
MDS error output
Shepard plots improved
Classical MDS allows placing origin on a specific point or on the data centroid (most usual)
Tow new similarity indices
French version available
Save discriminant analyses results
Save comparison results
Save other analyses results
Conversion to PA makes new matrix
Varimax rotation implemented in Principal Components Analysis
Minor corrections: Group selection dialog with order capabilities
Removing 2nd order confusion matrices, to avoid too many results in comparisons
Bug correction: 3D scatter plots bug
Bug correction: Null row checking in PCA
Null row/column advertising improved in CA/PCA
Improving results of MDS: Providing absolute relative and cumulative percentages of variance.
Improving matrix comparison dialog and results.
Improving results of discriminant analyses: Provinding confusion matrices for both resubstitution and leave-one-out procedures.
Version 1.4.7 - (07/Feb/2006)
Improvement of diversity analysis. More explicative output. Clarification of computations.
Fidelity analysis simplification: Elimination of building fidelity space
Bug correction: Exporting PCM cluster fuzzy sets.
Improvement of PCM clustering analysis
Saving temporary files of projects to avoid loosing information when writing error occur
Bug correction: Variable selection in covariance matrix dialog
Computing weighted euclidean distance
Bug correction: Centroid coordinates in MDS for fuzzy partitions
Output of intercluster distances in PCM
DBSCAN
Not allowing PCM when no starting classification is available
Bug correction: Linked object vectors in classification output
Bug correction: Shepard diagram in PCoA/NMDS removed because induces to erroneous interpretation
Allows to restrict the variables shown in biplot representations
Comparison of symmetric matrices by means of Gower distance (stress measure 1)
Output of one tailed p-values in Mantel test in matrix form
Version 1.4.3 - (31/May/2005)
PCM corrections
Defuzzification of FCM fuzzy partitions
Improvement of Normal Cluster Generation
Update of tree selection in Analysis Manager when selecting panels
Minor corrections
Linear discriminant bug correction: Selection of variables was not affecting eigenvalue decomposition
Adding new biplots and new output matrices (Z matrices) to CCA
Convergence of Canonical Populations Analysis and Linear discriminant Analysis
Inactivation of Canonical Populations Analysis
Bug correction in save hierarchical trees to overcome nesting limitation
Version 1.4.1 - (09/Nov/2004)
PCM switched to meta-PCM algorithm: automatic set of reference distances and distance bound tipicality correction
Variable selection in Linear canonical discriminant analysis
Minor corrections
Version 1.4.0 - (17/Sept/2004)
Bug correction: Creating number of clusters partition from hierarchical classification with reversals
Bug correction: Discrimininant analysis with training partition fitted to training data elements
Default saving enabled for all analyses (classifications and ordinations)
Bug correction: Import of XML relevé tables.
Imported taxon name is preferably the original name (to preserve uniqueness) but will be the accepted name if the former is an empty string. If both were empty
the taxon code would be used but taxon codes are not expected to be written in relevé files.
Pseudo-F statistic in variance cluster description
KM/FCM do not save original distance matrix for memory efficiency
Silhouettes computed in KM/FCM automatically after running partitioning algorithm (if distance matrix available)
Fuzzy Rand Index. Fuzzy confusion tablesDouble Partitions of objects and variables
Import from TWINSPAN outputs a double partition. Whole TWINSPAN output file text is recovered as analysis text too.
REBLOCK clustering method (Podani & Feoli 1991, Journal of Vegetation Science 2: 435-444)
Phi correlation in partition comparisons
Dendrogram improvement
Goodall probabilistic coefficient
Sociological space construction
Minor bug corrections
Version 1.3.6 - (28/May/2004)
Copy all matrix for export button
Order rows by label
Order rows by column value
Classification values used for colouring data tables
Right-click mouse functions in Analysis Manager tree and Data Editor tree
Version 1.3.5 - (06/May/2004)
Fuzzy Medoids and (manhattan) distance to medoids in cluster description
K-medians/FCMedians implemented: Centroids in partitions can be now medoids, so the generic name is 'prototypes'
Bug correction: Save prototypes in partitions when available
Bug correction: Comparison of partitions with unequal object lists and object order
Output p-values for correlations in variable covariantion analysis
First column made static in tables. This allows exploring data matrix without losing object reference
Improving of data matrix editor capabilities
Version 1.3.4 - (07/Apr/2004)
Bug correction in quadratic discriminant
Bug correction: save canonical population analysis
Math axes bounds can be changed in scatter, series, and bar plots
JCompas added as an 'other module'
Cluster generation prints output
Version 1.3.3 - (12/Mar/2004)
Translation of menu options
Save/restore language and Look&Feel
Automatic detection of language (spanish/english)
Program translation (to spanish and catalan)
Welcome dialog improved (all modules)
Version 1.3.2 - (09/Mar/2004)
Fidelity profiles available
Partition comparisons improved
Hierarchical comparisons improved (plots automatically generated)
Minor corrections: deleting data matrices in analysis manager
Silhouette plots implemented
New value separator in import: ';'
Checking of repeated variable or object labels. Done for ensuring data integrity.
Import/Export of CANOCO/DECORANA species files for exchanging data with programs CANOCO and PC-ORD
Import/Export of CANOCO environmental files
RDA/CCA minor corrections on error detection and error dialog prompting
Version 1.3.1 - (5/Feb/2004)
NEW MULTIVARIATE METHOD IMPLEMENTED: Canonical Population Analysis
Fowlkes-Malkows Bk index for classification comparisons
Comparisons of trees at different partitioning levels
Import TWINSPAN results in 'Analysis Manager': Two classification matrices are created, for species and sites respectively
Avoid duplicate labelings in data item names and analysis names
Menus in analysis manager reorganized
Key shortcuts for most used methods have been modified
Biplots with classification on object series
Version 1.3.0 - (8/Jan/2004)
Minor corrections on ginkgo project dialog
Default decimal format for data in project
Improvement of Precision Dialog with the oportunity to change decimal separator
Data values are from now on centered in data cells
Import/export can be done now using '.' or ',' as decimal separators
Version 1.2.9 - (19/Dec/2003)
Bar plots with more than one bar series
Plot frequency histograms for variables inside and outside a fuzzy set cluster
Print graphics with prompt dialog for user
Species diversity analysis improved
Antialiasing in plots can be disabled for printing or exporting
Prompt dialog for adding symmetric items when computing ressemblance indices...
Matrix correlation error dialog when trying to compute it using unequal size matrices
Project file in title
Data Import improvements
Version 1.2.8 - (17/Nov//2003)
Minor corrections
Import Releve Tables using User-defined Braun-Blanquet transforms
Export data matrix directly from Analysis Manager
Graphic Editor is now common to all Modules
Isolation for limit Cluster members
Plot frequency histograms for variables
3D Plots available
Plots made now using Monarch Charts 1.5 instead of 1.2 (copyright Singleton Labs ltd.)
Version 1.2.7 - (28/Oct/2003)
Isolation statistics in Cluster description
Out of memory error prompted as dialog
Separation of description of cluster in mathematical terms from species based information
Insert Row/Columns
Removing groups thread
Look&Feel changes (colors, icons...).
Version 1.2.6
Look and Feel changes
Version 1.2.5
Beta flexible clustering with beta variable
Cluster sphericity
Minor graphic corrections
Transform values of symmetric matrices
Splitting data into smaller subdata following a Partition
Removing groups from data
Merging two matrices as new ROWS or new COLUMNS
Version 1.2.4
Minor Corrections: Canonical Ordination output "%"
Thread priority.
Running analysis in red
Silhouettes in Cluster Description
Corrections on saving data
Minor corrections: Dendrogram to Partition
Version 1.2.3
Improvements of MINAR and Matrix Correlation
Miscelánea