L-5-The Fill From Example Tool

The Fill From Example Tool
The Fill From Example tool can be used to fill in the blanks in a data set by
populating the empty cells with the most probable value.
It can also be used in an iterative manner, where an expert partially
populates a column with new information, runs the Fill From Example tool,
refines the result, and runs the tool again until the result is satisfactory.
The Fill From Example Tool
Assume that a seasoned sales specialist (an expert) assigned flags (Yes or
No) to a few customers in the table, based on the specialist’s experience and
customer knowledge.
The Fill From Example tool learns, based on the flags assigned by the
specialist, and automatically fills the rest of the column with Yes and No
flags.
The Fill From Example Tool
Running the Tool and Interpreting the Results
To run the Fill From Example tool,
select the third button on the Analyze ribbon, Fill From Example
In the dialog box, you must
select the column containing
the partial information
entered by the specialists (in
this case, the High Value
Customer column).
The Fill From Example Tool
Running the Tool and Interpreting the Results
As part of the same heuristic analysis, the tool tries to detect the columns
that may be significant for the analysis.
This detection focuses mostly on finding and excluding those columns that seem to
contain unique identifiers, which are typically not significant for pattern analysis.
the specialist who partially filled the
target column may have ignored
certain columns that are, in his or
her experience, not significant.
The Fill From Example Tool
Running the Tool and Interpreting the Results
After the execution completes, two new elements are added to your workbook.
•First, a new column is added to your table, containing the auto-fill results produced
by the tool
The Fill From Example Tool
Running the Tool and Interpreting the Results
• Second, a new spreadsheet is added to your workbook, summarizing the patterns
detected by the tool and used for auto-filling.
The patterns report explains the rules detected by the tool by analyzing the non-empty
cells in the original High Value Customer column
the report shown in
Figure shows that
European customers
with a commute
distance of 2 to 5 Miles
and 5 children are
generally considered
high-value customers.
The Fill From Example Tool
Refining the Results
If the new values in High Value Customer Extended are what you expected (and
the patterns are reasonable), you are probably done with the Fill From Example
tool. However, often this is not the case.
The set of hint values in the original High Value Customer may be too small, or
the patterns may not reflect your expectations.
In either of these cases, the results of the tool may be refined.
The Fill From Example Tool
Refining the Results
For example, based on your experience, assume that the customer with 1 car and
a commute distance of less than 1 mile, at row 18 in the data set, is not a highvalue customer, although the tool signaled it as being one.
Modify the High Value Customer column in the spreadsheet by replacing the blank
with your expert knowledge, as shown in the Figure. Then, run the Fill From
Example tool one more time.