`<retail:basket_affinity>`

<retail.basket_affinity> expands upon <affinity> and supports the common specific retail use case of calculating basket affinity.

Description

<retail:basket_affinity> makes the more general <affinity> operation convenient and specific for retail use cases by performing some additional calculations after the operation, by giving attributes more retail-specific names, and by providing a TRS panel for the operation.

Syntax

<retail:basket_affinity
         basket_id="[TRANSACTION_ID_COLUMN]"
         prod_col1="[PRIMARY_PRODUCT_COLUMN]"
         prod_col2="[SECONDARY_PRODUCT_COLUMN]"
         prod_filter1="[PRIMARY_PRODUCT_VALUES]"
         prod_filter2="[SECONDARY_PRODUCT_VALUES]"
         copurchase_threshold=
         avgcol="[SALES_COLUMN]"
         ignoresegby="0|1"
         removediag="0|1"
         removebelowdiag="0|1">
</retail:basket_affinity>

Attributes

basket_id: The "basket", or transaction ID column. The table must be segby this column, unless you indicate ignoresegby="1". This attribute is required.
prod_col1: The primary product column for comparison. This attribute is required.
prod_col2: The secondary product column for comparison. This attribute is required.
prod_filter1: An optional filter for the primary product values in prod_col1.
prod_filter2: An optional filter for the secondary product values in prod_col2.
copurchase_threshold: The default value is "0.0".
avgcol: The sales column. This attribute is optional.
ignoresegby: Set to "1" to ignore the segmentation requirement. You may return nonsensical results if the data is not in fact segby. The default value is "0".

Example: Retail affinity analysis

The following example demonstrates how to perform affinity analysis on products of the group Snack-Chips/Pretzels. We set the prod_filter1 attribute to "1460" (for the chips/pretzel group). We limit the results so that there are a minimum of 100 transactions containing both items: the chips/pretzel item and the accompanying item ("coinc>=100"). Then we sort in descending order by average basket, so we can see which items when purchased with a chips and pretzels item resulted in the largest baskets.

<base table="retaildemo.retail.sales_detail"/>
<link table2="retaildemo.retail.products" col="sku" cols="group"/>
<retail:basket_affinity basket_id="transid" prod_col1="group" prod_col2="sku" prod_filter1="1460" 
    avgcol="xsales"/>
<sel value="coinc>=100"/>
<sort col="avgbskt" dir="down"/>
<link table2="retaildemo.retail.products" col="key1" col2="group" cols="groupdesc"/>
<link table2="retaildemo.retail.products" col="key2" col2="sku" cols="description"/>
<col name="groupdesc" fixed="1"/>
<col name="description" label="Description" fixed="1"/>