What is Assisted Analytics?
Assisted Analytics (AA) is an optional features for a Measure Factory implementation. When properly licensed, it provides the following:
- Additional configuration options for the factory in Workbench
- Additional options for the Analysis portlet in DivePort
- Additional help to analysts examining their data in Measure factory
Assisted Analytics adds a new mode to the Analysis Portlet, which performs automatic dives and summarizes the results, to give users ideas about where to go next in their data analysis. In those dives, dimensions and dimension values are examined, and given a "rating".
Why Assisted Analytics?
Consider the analysis possible from a typical Measure Factory dashboard:
Without pre-configured dive paths or hierarchies, the diving options are many. As an example, if there are 7 measures, 69 dimensions, and 51 different suppliers, and you make only a three level dive, there are over 100 million possibilities. Even though it is completely automated, it can be overwhelming. Also, manual diving takes time and important causes behind the numbers can be missed.
With assisted analytics, you can let the software help you find the interesting data. It can look ahead, rate the current data, and give suggestions. These features are implemented as options available in the DivePort Analysis portlet.
In its current form, the automatic dives look at how each dimension value compares to its peers in each of the last 12 months, using the current measure. If in one month, a dimension value is more than 3 standard deviations away from the mean for that dimension, then it is considered an outlier. The overall rating for a dimension value is the sum of weighted points for each month that it is an outlier, normalized to a 1-9 range. The points are weighted so that more recent outliers will tend to have a higher rating than less recent ones.
Dimensions eligible for AA are configured in Workbench and made available in DivePort. Once the portal administrator defines the Analysis portlet, the Analyst will have access to lists of dimensions and values by an importance rating.
See also: