Visualize the Data
There are two visuals that may help to determine which characteristics to include in the analysis as well as which Data Type to use for each characteristic. The first appears before selecting a Target field by clicking on the name of any characteristic. By doing this, the percentage of records for up to ten of the unique values or bins of that characteristic will be displayed to the right of the table, as shown below.
The second visual is a graph which appears in place of the percentages above after selecting a Target field, again by clicking on the name of a characteristic. Along the x-axis are bars for up to ten of each bin or value. The width of each bar represents the proportion of the Exposure in that bin or for that value to the total exposure for that characteristic; if no exposure is selected, an exposure of 1 is assumed for each record. The y-axis shows the Total Target divided by the Total Exposure in each bin or for each value. Since the width of each bin is the Exposure, and the height of each bin is the Target/Exposure, the area of each rectangle represents the Total Target for each bin or value.
Also shown in the graph is a red line which represents the Total Target divided by the Total Exposure across all bins or values. This line can be thought of as the Base Factor. Bins or values whose bars extend above this line will have a raw factor greater than 1, and bins or values whose bars fall below it will have a raw factor less than 1. The height of each bar divided by this Base Factor can be thought of as the raw factor for the corresponding bin or value. See an example of the graph below.
To determine whether to interpret a characteristic's values as either unique (Generic Characteristic) or in bins (Grouped) for displaying the visuals described above, MultiRate follows a set criteria. If the data is numeric and the number of unique values for the given characteristic is greater than the minimum of 50,000 and 20% of the total records (see formula below), MultiRate automatically groups the characteristic into 10 bins, each containing roughly equal exposure. All other characteristics are treated as Generic Characteristics. In this case, if there are more than 10 distinct values, the 9 most significant values are displayed and all other values are grouped into an "All Other" bin. The actual Data Type selected by the user and/or user-defined bins are not used in these calculations.
Criteria for Grouped Characteristic:
Each of the visuals described above may be printed via the "Print" button.