+

Search Tips   |   Advanced Search

Configure the Item Affinity Engine

For every cross-selling transaction in the user's shopping history, the Item Affinity Engine derives its calculations from the following statistics:

In general terms, we use the Item Affinity Engine as follows:

  1. Define the shopping cart analysis requirements.

    This refers to the scope and filters for predictions you plan to use for the shopping cart analysis. This definition is very flexible.

    For example, the shopping cart can be based on the user's entire shopping history, on seasonal product sales, on the user's current session, user page views critical to an item's purchase, and so on.

    The type of purchases you expect from the users will determine the type of shopping basket we define.

    For example, a session-based shopping cart would be suitable for a grocery store site, where users tend to purchase a variety of items at once. A lifetime shopping cart is suitable for a site where users tend to purchase one large item occasionally, such as expensive electronic equipment.

  2. Configure an item affinity set for the shopping cart analysis.

    The item affinity set defines the shopping cart analysis requirements. Similar to mentor sets or transaction sets, the item affinity set defines the type of data to collect. The item affinity set starts by collecting transactions from a transactions input table similar to Lps_User_Trx. We can include filters to query fields in the input table, specify the types of transactions to query, the number of highly associated items to consider, and so on, for the shopping cart definition. Finally, we specify an output table to which the results of the analysis are written. The output table must have the same fields and datatypes as the Lps_MBA_Scored table.

  3. Run the accumulator utility to collect the item affinity set data and populate it into the output table specified in the item affinity set.

    Similar to the sifter, which is used for the other LikeMinds server engines, the accumulator uses the item affinity set data to make the support, prediction, and confidence calculations (described earlier). It writes its findings to the output table specified in the item affinity set.

    You should configure the accumulator to run at times of low database usage, since it can make heavy use of system resources.


Parent: Configure the LikeMinds engines