Insightful Miner 8 Fixed Bugs IMPORTING AND EXPORTING SPSS 12 can handle mixed case variable names, so importing SPSS files no longer automatically converts column names to lowercase. [40590] When a "Read Sybase - Native" node is used, the Sybase connection library is left loaded for the duration of the session. Previously the library was unloaded between requests, which could cause failures on Windows. [40678] The file name is included in the error message when a Read File node fails because of an impropper file name. [40849] Allow ascii characters \200-\377 (pound sign, O-umlaut, etc) to be specified in Read/Write database nodes (in the table names, SQL string, etc), and in the delimiter string for the Read Text node. [40944, 41243] IM hexbin matrix node now correctly scales gray values. [40946] The Read Text File Node correctly implements the Start Row value. [41226] I-Miner now writes an to an Excel file a data set that is 59009 by 10. Previously, a file of that size was corrupted. [42298] The Write Excel Node now generates an error if the specified file name is in a non-existent directory. When the dialog for this node is closed, a warning window is shown if the file name refers to a non-existant directory. [43759] GRAPHS AND CHARTS All of the S-PLUS plotting nodes (Density plot, Hexbin plot, etc) now work correctly when plotting data columns that are not legal S-PLUS names, such as "A_A" or "B!". The column names shown in drop-down lists in these node dialogs are not modified to make them S-PLUS legal. [40868] In Insightful Miner 7, a surface plot with a 40x40 grid would return an error reading XML. This problem is fixed in Insightful Miner 8. [40936] Hexbin Plot now correctly uses the Max Rows value to specify the number of points to sample. [41251] hist node now recognizes columns that where originally passed with underscores in the column names when the All rows checkbox was checked under properties. [41526] DATA MANIPULATION AND TRANSFORMATION The Sort node will clean up temp files if it fails because the disk is full. [31465] It is no longer necessary to open the Append node property dialog to get it to recognize when more than two inputs are specified. [40751] If the expression language function random() is called multiple times within a given node, for example if it is used in two column expressions in the Create Columns node, then the random numbers produced by the multiple calls are independent. [40895] Text to numeric conversion in Expression Language now correctly converts "+5" to positive 5, in addition to converting "-5" to minus 5. [41022] Sort no longer compares strings past the end of the string. [41081] The error message generated when Sort has no data is clearer. [41214] The new names will be saved/restored when the Save/Load buttons are used in the Modify Columns dialog and tab. [41358] Fixed a bug where the Join node could output incorrect results if multiple join keys were specified in a different order than the columns in the input datasets. [41709] Fixed a bug where the Unstack node could output incorrect wrong values if a group column value is the same as a key column name. Now it will print a warning in this case, and rename the key column to avoid conflicts. [42167] Using categorical columns as key columns for the Unstack node no longer results in empty key columns in the output. [44382] MODELS Ensemble trees saved as PMML and imported back into Insightful Miner are restored correctly. [41170] The Predict node no longer runs indefinitely when the input has no rows. [42946] Naive Bayes will output PREDICT.class when the user has set the appropriate property. [43564] Fixed a bug where the Predict node could in rare circumstances give an error "number of rows in the output buffer XXX is less that the number of rows in the input buffer YYY" if no-caching was specified. [44228] S-PLUS SCRIPT Problems with column name mangling in the S-PLUS Script node have been fixed. See the Release Notes for more information. [40769] If an S-PLUS Script node has a script that outputs a data.frame containing a column of complex numbers, a warning is printed stating that the imaginary parts of the numbers are discarded. [40789] The S-PLUS Script node with "Execute BigData Script" box checked will always pass the in1.column.roles information to the script function. [40819] When an S-PLUS Script node is executed with "Execute Bigdata Script" enabled, it is executed as if "dynamic.outputs=T" was output during the test phase. Whatever column names, types, roles, etc were output during the test phase (if there was one) will be overridden by the value output by the bigdata script. Therefore, if you wish to specify output column roles for the first output, they will have to be output as the "out1.column.roles" element of the output list when the output is generated. [40829] Fixed a bug where the S-PLUS Script node in1.column.min and in1.column.max values for timeDate columns are incorrect. [41573] In an S-PLUS Script node, when using dynamic.outputs=T in the Test Phase, column names now change consistently. Previously, if the input contained an underscore, the name would change in the output, with underscores replaced by dots. With dynamic.outputs=F, the underscores did not change. [42525] VIEWERS The Neural Network Runtime Viewer button is no longer permanently disabled. [40821] The XSL transform engine supports processing of larger XML files. [40883] In a multiple-output node such as Partition or Split, the output viewed in the Data Viewer can now be changed when any tab is displayed. Previously the output could only be changed when the "Data View" was displayed. [40927] The "No data" message box is no longer displayed when the Data Viewer is opened on a data set with no rows. [40971] If Insightful Miner's install directory is read-only, previous versions of Insightful Miner would fail to display HTML. [41541] Fixed several bugs converting numbers to strings or formatting numbers in the viewer when the "Decimal & Thousands Symbols" option is not the default (decimal==, thousands separator=="). [44348] STARTUP The "IMiner" script on Solaris sets JAVA_HOME to use the JRE included in Insightful Miner. [41006] The memory/heap settings have been raised to a heap max of 500m and stack frame size of 1280k. [44267] DRAG AND DROP Drag and drop from the Worksheet to the Explorer Pane has been improved. [40965] COMMENT EDITOR The Comment Editor has a minimum size and can be resized. [24532] The Comment Editor has word wrap and scroll bars. [40807]