Profiler, ADS, and TWM - Teradata Warehouse Miner

Teradata® Warehouse Miner™ Release Definition

Product
Teradata Warehouse Miner
Release Number
5.4.6
Published
November 2018
Language
English (United States)
Last Update
2018-12-07
dita:mapPath
erc1538428083343.ditamap
dita:ditavalPath
ft:empty
dita:id
erc1538428083343
Product Category
Teradata Tools and Utilities
Reference Number Description
TWM-423 Decision Tree graph and rules do not display dates in local format.
TWM-710 A series of Redo operations after series of Undo's may jump to end state unexpectedly, then resume normally.
TWM-726 The results viewer does not in most cases display complex data types such as geospatial or temporal data, but rather requires that such data be cast to a character data type, if possible.
TWM-835 When a user attempts to uninstall TWM while the program is running, the Windows uninstall program presents a dialog box to explain that the current program must be closed first. However, the list in this dialog that should name the running program is blank. The user should choose the “Automatically close applications…” radio button and proceed.
TWM-848 The Delete key does not always work in the Free Form SQL editor due to format changes. The workaround is to highlight the text to delete first.
TWM-1064 Metadata timestamps are based on local client time rather than database times.
TWM-1082 When installing any TWM product, installation errors occur if auto-run installation option is selected from a CD. The workaround is to open the Windows Explorer displaying the CD and execute the .msi file from there, or if need to be administrator, execute .msi file from a command prompt opened as administrator.
TWM-1329 Tree Scoring Node – when the ‘Evaluate’ or ‘Evaluate and Score’ options are used with new data, the output is not correct. The lift charts that are generated when scoring and evaluating new data do not assign the results properly into deciles. This results in charts and tables that do not provide useful insights.
TWM-1475 When trying to build a decision tree on more than 1000 independent variables the decision tree fails right away in initialization. Workaround: Reduce the number of variables you are selecting to below 1000 or use the Gain Ratio Extreme method.
TWM-1492

Scatter Plot returns values out of range for Aster. Later versions of the Aster ODBC driver (6.10 and 6.20) exhibit this error whereas an earlier driver (5.10) does not. The issue has to do with queries against the Aster NUMERIC data type which does not specify the size of the numeric values. This can affect other graphs that have the drill down feature when you try to drill down and a NUMERIC field is included.

There are several ways to work around this problem. Make sure any data being sent to the scatter plot or drill down of histogram, values, statistics or histogram graphs do not use a NUMERIC type but are cast to FLOAT or NUMERIC(X,Y) instead. For instance, create a view on a table that has NUMERIC fields and cast them as FLOAT.

TWM-1593
When Gain Ratio Extreme is selected, the following options should be disabled or limited:
  • Maximum Nodes label and text box should be disabled
  • Chaid Significance Levels label and Merging/Splittinglabel and text boxes (these are all already disabled)
  • Include Validation Table check box plus Validation table label and pull-down list should be disabled
  • Include Lift Table check box plus Response Values label, text box and Values button should be disabled
  • Pruning Method can only be Gain Ratio or (None) and should be limited
  • Gini Test Table label and pull-down list should be disabled
Expert Option about in-memory processing should be disabled (both labels and text box).
TWM-1697 The project Aster Tutorials - SQL-MR Examples has several failures when using Aster Analytics 6.21 or later due to building the project with an earlier release.
TWM-1718 Classic Histogram Overlay and Stats options reduce Overlay and Stats columns to 24 characters so may not be unique (theoretical problem).
TWM-1719

Decision Tree and Decision Tree Scoring are limited to input tables with approximately 2 billion rows. If this is exceeded, an error is given and the analysis terminates. The message is:

“Unable to build/score tree. Number of rows exceeds 2**31 (2147483648) maximum limit”

TWM-1720

The following types of analysis, when run against billions of rows of input data, may be impractical due to excessive run times or spool file usage.

  • Values analysis with count of unique values requested
  • Statistics analysis with extended options
  • Histogram with quantiles option
  • Histogram with overlay and stats
  • Adaptive Histogram with quantiles option
  • Scatter Plot (long load time such as 15 min.)
  • Variable Transformation Bincoding with quantiles option
  • Logistic Regression/Scoring Reports (lift, etc.)
  • Cluster and Cluster Scoring (all varieties except In-Database Fast Kmeans which is OK)
  • Various Statistical Tests
    • Median Test
    • 2-Way F-Test with Unequal Cell Counts
    • F-Test N-Way
    • Kolmogorov-Smirnov Test
    • Lilliefors Test
    • ShapiroWilk Test
    • D’Agostino and Pearson Test
    • Smirnov Test