|
||||||
|
|
||||||
|
||||||
|
Screenshots, continued 4. Single-variable tables For single-variable tables, you just select the variables in which you're interested, and where appropriate, choose rounding options in the right-hand panel. Numbers can be rounded to the nearest sensible size - say nearest hundred. Dates can be rounded to month, quarter, year or decade. Text variables can be truncated to their leftmost few characters. Where these rounding options are insufficient, then you can transform variables using the option on the previous form. (There's more detail on this on the next page.) Often, you will have a lookup table with descriptions of various codes - by choosing the Lookup option, you can easily specify where the descriptive values should come from. Another useful option is to combine small categories - often, you may have a variable with a large number of possible values, many of which occur rarely. You can choose to have all the rare values grouped into a single "other" category. (You choose how rare is rare.) 5. Cross-tabs In cross-tabs you also have to select a column variable. This analysis method is best used for comparative profiling of different subsets within the data; percentages and index scores are calculated to highlight the differences between subsets. As above, you can choose to show charts (stacked bar charts, which can of course be manipulated later in Microsoft Excel), you can highlight high or low index scores, and you can combine small categories into a single "other" category. Variable rounding and lookups are again available, too. 6. Response tables Here, you choose a single variable to be your response variable, and specify which of its values is the positive value. Then select the profiling variables you are interested in. As before, rounding and lookup options are available for the profiling variables, and you can opt to have charts, extra scores, and also to combine small categories. 7. Decision trees As above, you choose a response variable and its positive value, and the the candidate variables for inclusion in the tree analysis. You can also choose various options controlling the tree analysis, and whether to include a cumulative gains chart (as well as a cell-by-cell gains chart.) |
||||||
|
Screenshots | Sample output | Download | Purchase |System requirements | Feedback |
||||||
|
All content copyright ©2005 Andrew Gilham. All rights
reserved. |
||||||