Mirror of Oracle documentation
Converted for search and offline reading. Authoritative source: Oracle. Diagrams and some complex tables are simplified — check the PDF when in doubt.
7 Affinity Analysis
This chapter describes the use of the Affinity Analysis (AA) Cloud Service module.
Introduction
Market basket analysis involves the use of data mining techniques to search for sales patterns between products within a given group of transactions. The output of that analysis provides a rule that defines the association found between products at the subclass or class level of the merchandise hierarchy.
A rule consists of one to three antecedents (IF attributes) and a single consequent (THEN attribute). For example:
IF (milk) and (juice), THEN (cereal)
In other words, if a customer purchases an item from the subclasses milk and juice, the customer will also purchase an item from subclass cereal. After a rule is defined, a user can use the AA interface to understand how strong the affinity is, using rule confidence and support. The probability that a customer will buy milk, juice, and cereal is known as the support percentage , while the conditional probability that they will buy cereal when they buy milk and juice is known as confidence . Rules with a very high support value occur frequently in your transaction history, while rules with a high confidence value represent a strong affinity between products.
After users have identified selling patterns, they can begin to take action based on those patterns, as well as the needs and goals of their product category. Suppose that a merchant is tasked with bringing in more margin dollars to the cereal category. Using the affinity rule in the preceding example, the merchant might work with the dairy category on a milk promotion to increase sales of milk. This in turn increases the sales of cereal, without sacrificing margin dollars on a cereal promotion. Note that this can require cross-category planning in some cases, as product affinities can sometimes occur between seemingly unrelated products (such as pet food and beer).
Another component of market basket analysis relates to the product assortments being sold in stores. Using product affinities and sales history, AA provides assortment recommendations that improve the revenue or margin of a category by suggesting product additions or removals. Products may be recommended for removal if they are found to be too similar to other products in the assortment (and thus cannibalize the sales of those products). Conversely, products that are not similar to any other items in the assortment may be candidates for inclusion, as they will not divert sales from the existing assortment.
Market basket rules are used to improve the assortment recommendation by showing the potential lift (or halo effect) on your overall sales due to any known affinities on recommended item additions. For example, if AA is analyzing an assortment for Coffee, and a particular item is part of a market basket rule that drive sales for Milk, then that item has a greater potential value for the lift it brings to the Milk category. AA may then recommend that item over other items in the category, because including it will bring in additional revenue to other assortments without changing those assortments directly.
User Interface
The AA user interface consists of several screens that help the user analyze and take action on the results of the market basket analysis and assortment recommendation processes. The following list summarizes the main purpose of each screen:
-
Assortment Recommender - Review the product assortment recommendations made by AA, including the expected halo effects from items in the optimized assortment.
-
Top N Class Affinities - Review the market basket affinity rules identified by AA between different product classes, including insights around the frequency and profitability of the top rules.
-
Top N Subclass Affinities - Review the market basket affinity rules identified by AA between different product subclasses, including insights around the frequency and profitability of the top rules.
-
Top N Promotion Affinities - Review the market basket affinity rules identified by AA between different product subclasses under the effects of a promotion, which can be used to identify the effects promoting a category has on other non-promoted products.
-
Top N Customer Segment Affinities - Review the market basket affinity rules identified by AA between different product subclasses under the effects of a promotion and separated by customer segment, which provides insights into the top promoted product affinities for a targeted group of customers.
The way that you interact with the AA user interface depends on your business role and which insights you want to take action on. For example, a category planner looking to make changes to their assortment to increase margin dollars might start from the Assortment Rec screen. The planner might select the category and one of the top-selling locations, and then review the added or dropped items AA has recommended. From there, the planner can choose to move forward with the recommendations and make the necessary adjustments to the assortment plan or dive deeper into the Product Affinities screens to better understand why certain recommendations are being made.
Regardless of which screen you are examining, a set of global prompts determines the data displayed throughout the application.
On first accessing the application, you make selections from the prompts and click on the Submit button, which will load the remaining application screens based on your selections. Modifying and submitting the prompt values after that time will refresh the screens with new data.
Table 7-1 AA Global Prompts
| Prompt Name | Description |
|---|---|
| Top 10 By | Select the primary metric used to rank the market basket results in all places where affinity rules are displayed. |
| Fiscal Week From | Select the starting week from which results must be displayed. The values displayed are the week of the Fiscal Calendar. |
| To | Select the ending week that results must be displayed for. |
| Department | Select the department from the merchandise hierarchy that results must be displayed for. All of the classes or subclasses on the THEN side of an affinity rule will be from the selected Department. |
| Location | Select the location from the location hierarchy that results should be displayed for. |
The first step in using the assortment recommender screen is to select a store and assortment to review. The stores listed in the drop-down menu are limited to those stores that have completed an assortment recommendation calculation for the selected department in the selected period. The assortment drop-down menu is limited to those assortments with a completed assortment recommendation calculation for the date range, department, and location previously selected, and align with the level of the product hierarchy that assortmentplanning operations occur at. For example, if we process Coffee on Week 1, Yogurt on Week 2, and Milk on Week 3, and we pick a date range of Weeks 1-3 in the global prompts, we will show results for those three categories.
After selecting a store and an assortment, the screen displays data showing the systemoptimized product list and the original product list, along with several summary metrics.
Table 7-5 Assortment Optimization Summary Metrics
| Field Name | Description |
|---|---|
| SKUs Added | The number of SKUs added to the assortment by the optimization process. |
| SKUs Dropped | The number of SKUs removed from the assortment by the optimization process. |
| Amount % Variance | The change in sales retail amount for the assortment after SKUs are added and dropped by the optimization, based on average weekly sales. |
| Units % Variance | The change in sales units for the assortment after SKUs are added and dropped by the optimization, based on average weekly sales. |
The following table lists the summary metrics that the optimized assortment table displays insights into expected product performance if the recommendations are applied to your assortment.
Table 7-6 Optimized Assortment Table
| Field Name | Description |
|---|---|
| Item | The item number and description of the SKU in the assortment. |
| Description | The recommended change for a SKU in the assortment. This may either be to keep an existing SKU or add a SKU that was not previously in the assortment. |
| Base Sales Units | The average weekly sales units of the SKU at the selected location. |
| Halo Sales Units | The average weekly sales units of the SKU at the selected location, after adjusting for demand as a result of Halo sales due to market basket affinity rules involving the SKU. |
| Base Sales Amount | The average weekly sales amount of the SKU at the selected location. |
| Halo Sales Amount | The average weekly sales amount of the SKU at the selected location, after adjusting for demand as a result of Halo sales due to market basket affinity rules involving the SKU. |
| Base Sales Profit Amount | The average weekly sales profit of the SKU at the selected location. |
| Halo Sales Profit Amount | The average weekly sales profit of the SKU at the selected location, after adjusting for demand as a result of Halo sales due to market basket affinity rules involving the SKU. |
| Incremental Demand | The number of sales units of the SKU that do not transfer to any other SKU in the assortment if it were to be deleted from the assortment. |
ee
SCC‘
|
2. After AA has been provided with the assortments and lists of possible changes that Anne wants to make, the system will schedule and execute the optimization process using predefined business rules (such as the maximum number of items that can be added or dropped by the system).
3. Knowing the process executed over the weekend, Anne logs into AA on Monday morning to review the results. She selects a location and her Coffee assortment in the UI prompts, and then begins to analyze the recommendations.
4. AA has selected three items in the current assortment to be dropped and another four items that should be added. Anne first looks at the dropped items and compares them to the other items she was thinking about removing. She notes that the dropped items are not the worst-selling ones she had chosen, but AA shows them as contributing very little to halo effects in other areas, so they may have very weak market basket affinities that contribute to their removal.
5. Anne next reviews the recommendation for items to add to the assortment. Of the four items that AA has chosen, three of them have significant halo effects driven by the market basket affinities. Anne is not as sure about the last item, so she selects it and clicks on the Cannibalization button. She notes that the item has relatively low substitutable demand from other items, suggesting it is not very similar to anything in her current assortment. This makes it a good candidate for addition, as it can bring in new demand that the assortment may not have today.
6. Anne decides to accept all of the recommendations made by AA for this assortment, exports the results to Excel for later reference, and then exits the system to make the necessary changes to her assortment plan.
Market Basket Analysis
The market basket analysis screens of the application all relate to the market basket affinity rules generated by AA. The tiles and screens provide a way to view different sets of data, depending on the configurations used for your business.
Market basket rules and their associated metrics are a key insight into product demand and customer buying behaviors. Understanding which products are more likely to sell together allow your business to coordinate targeted offers and promotions for products that have strong affinities elsewhere, without having to promote those other categories. It is also possible to understand market basket differences between customer segments, allowing you to further refine your sales and promotion strategies to target specific groups of customers that will bring in the most value to your business.
There are five screens available for viewing the AA results, each containing a different subset of affinity rules with certain characteristics as described below.
Table 7-10 Market Basket
| Field Name | Description |
|---|---|
| Class | Displays market basket affinities between different classes, without any consideration for whether the items were on promotion or not. |
| Subclass | Displays market basket affinities between different subclasses, without any consideration for whether the items were on promotion or not. |
| Promotion | Displays market basket affinities between different subclasses when the antecedent (if) components are promoted and the consequent (then) components are not promoted. |
|
Table 7-15 (Cont.) Targeted Product Affinities Detail Metrics
| Field Name | Description |
|---|---|
| Avg Sales | The average weekly sales retail amount for the IF and THEN items in the rules. |
| Avg Profit | The average weekly sales profit for the IF and THEN items in the rules. |
| Avg Quantity | The average weekly sales units for the IF and THEN items in the rules. |
| Baskets | The number of market baskets that contain items from the IF and THEN components of the rule. |
| Support | The percent of market baskets that contain items from the IF and THEN components of the rule. |
| Confidence | The ratio of the number of transactions where the entire rule is present, compared to all transactions containing IF items. |
| Affinity Reverse Confidence | The ratio of the number of transactions where the entire rule is present, compared to all transactions containing THEN items. |
| Lift | A statistical measure of how strong the product affinity is, as compared to random chance for the rule (higher is better). |
| % MB Sales | The percent of total sales retail amount that the rule represents. |
| Target Sales Value | The average weekly total sales retail amount across all transactions in the selected periods. |
| Target Profit | The average weekly total sales profit across all transactions in the selected periods. |
| Target Sales Quantity | The average weekly total sales units across all transactions in the selected periods. |
This view of market basket affinities provides powerful insight into the relationships between different product categories, as it may expose affinities even between seemingly unrelated categories that your customers tend to purchase together. This screen allows you to see how actions taken on the target (IF) products will impact other areas within your business. For example, any action taken to increase the sales of an IF category can lift the sales of all other products with a strong affinity to those items. Conversely, if you have a category that is performing poorly on the THEN side of a rule, you may be able to take action on the IF products to boost that category’s sales indirectly.
Using Market Basket Rules
There are a variety of ways to take action within your business based on the insights that AA provides. The most common use is to develop a promotion strategy that takes advantage of product affinities to maximize halo effects across other categories. By reducing the amount spent on promotions while increasing the effect of those promotions, you can realize additional revenue and boost profit margins. For example, by analyzing the AA results, it is possible to identify multiple product categories with affinities on a target category, where some categories are more profitable than others. This can provide alternatives for promoting items that will yield more in sales or profit than a more obvious affinity like Hot Dogs and Hot Dog Buns.
Another common practice is for store planners and planogrammers to take advantage of product affinities when deciding how to arrange the products on the shelf or for adding aisle end-caps for strongly associated product categories. AA can show obvious relationships like Bread and Peanut Butter, but it may also reveal previously unknown associations like Pet Food and Beer. Using market basket rules to inform store layouts allows you to place commonly
purchased groups of products close together in the store, increasing the chance of customers purchasing more from those categories.
Combining market basket analysis with customer segments further enhances the potential benefits of AA. Customer behavior information is obtained from mining transaction history, and it is correlated with customer segment attributes to inform your promotion strategies. The ability to understand market basket affinities allows marketers to calculate, monitor, and build promotion strategies based on critical metrics such as customer profitability and preferred categories.
As an example, consider a marketer that is planning promotions for the Soda and Chips categories. While analyzing the data in Affinity Analysis, the marketer finds that a particular customer segment, College Singles, has a very strong affinity for buying Soda any time they purchase Chips.Armed with this knowledge, the marketer may decide to create a targeted offer for this customer segment only on the Chips category, knowing that they are likely to also buy Soda in the same basket. This kind of offer is more cost effective than a company-wide promotion on both the chips and soda categories, but may yield almost the same results, because the offer has focused in on the customers that will generate the most revenue in these categories.
Affinity Analysis
This section addresses MBI analysis. The functionality described here includes creating custom runs and comparing the results of multiple runs.
System runs are executed automatically once a week as part of the weekly batch cycle using settings that have been configured for a specific level of the location hierarchy. Since system runs occur weekly, they can be used to examine results over a long period of time.
User runs can be run at any time by an individual user who specifies the settings for the run.
Two types of comparison are available: Compare Runs and Compare Results.
Use Compare Runs to select specific runs to compare. This comparison is useful for straightforward runs such as comparing the results of two different months.
Use Compare Results to compare system runs only, with specified dates and other filters. In this case, take care in selecting runs that overlap, as some runs may be represented in the results more than once.
Affinity Analysis Overview
The Affinity Analysis Overview screen provides the tools in the following images. In addition to the standard application functionality, you can also
-
Select from the following functionality using the Action drop-down menu:
-
Create Custom run
-
Delete
-
Refresh
-
Add to Baseline Set
-
Add to Comparison Set
-
-
Select the type of run to view in this screen: User-Run, System-Run, or all runs.
SCC‘
The following table describes the options you can use to filter the results that will be displayed.
Table 7-17 Filter Results
| Field | Description |
|---|---|
| Department | Select one or more departments for which the results should be filtered. |
| Item | Select the item within the department from the drop-down list. |
| All | Select to prevent any other filters from being used. |
| Class | Select one or more classes for which the results should be filtered. |
| Top Results By | Select a metric to see the top results by: Net Sales, Net Quantity, of Net Profit. |
| Sub Class | Select one or more subclasses for which the results should be filtered. |
| Top Results Displayed | Select a value (10, 20, 30…970, 980) to limit the number of results displayed. |
Affinities Search Results
The Affinities Search Results section of the pop-up provides details about the observed associations (affinities) between sales transactions. It is expressed as “If a customer buys X, then that customer may also buy Y.” The table describes the relevant values associated with the If/Then transactions and contains the following fields:
Table 7-18 Affinities Search Results
| Field | Description |
|---|---|
| Customer Segment | (Customer Segment screen only) Displays the customer segment for which the product affinity was found. |
| Promotion | (All affinities screen only) Displays whether the rule represents IF components that were on promotion. |
| If | Displays the classes or subclasses containing items that were purchased in the market baskets for a given product affinity. All market baskets that contributed to a rule will contain items from these product categories. |
| Then | Displays the class or subclass that was found to have an affinity with the IF components of the market basket rule. |
| Baskets | The number of market baskets that contain items from both the IF and THEN components of the rule. |
| Avg Sales from “If” Items | The average sales retail amount from items belonging to the IF classes or subclasses. |
| Avg Profit from “If” Items | The average sales profit from items belonging to the IF classes or subclasses. |
| Avg Sales Quantity from “If” Items | The average sales units from items belonging to the IF classes or subclasses. |
| Avg Sales from Affinity Items | The average sales retail amount from items belonging to the THEN class or subclass. |
| Avg Profit from Affinity Items | The average sales profit from items belonging to the THEN class or subclass. |
| Avg Sales Quantity from Affinity Items | The average sales units from items belonging to the THEN class or subclass. |
|
Table 7-19 Affinities For
| Field | Description |
|---|---|
| Avg Number of Baskets | The average number of baskets across the runs for the combination of the If and Then items. |
| Avg Sales from “If” Items | The average sales retail amount from items belonging to the IF classes or subclasses. |
| Avg Profit from “If” Items | The average sales profit from items belonging to the IF classes or subclasses. |
| Avg Sales from Affinity Item | The average sales retail amount from item belonging to the THEN class or subclass. |
| Avg Profit from Affinity Item | The average sales retail amount from item belonging to the THEN class or subclass. |
| Then | Displays the class or subclass that was found to have an affinity with the IF components of the market basket rule. |
| Avg Sales | The average weekly sales retail amount for the IF and THEN items in the rules. |
| Avg Profit | The average weekly sales profit for the IF and THEN items in the rules. |
| Avg Quantity | The average weekly sales units for the IF and THEN items in the rules. |
| Baskets | The number of market baskets that contain items from both the IF and THEN components of the rule. |
| Support | The percent of market baskets that contain items from the IF and THEN components of the rule. |
| Confidence | The ratio of the number of transactions where the entire rule is present, compared to all transactions containing IF items. |
| Affinity Reverse Confidence | The ratio of the number of transactions where the entire rule is present, compared to all transactions containing THEN items. |
| Lift | A statistical measure of how strong the product affinity is, as compared to random chance for the rule (higher is better). |
| % MB Sales | The percent of total sales retail amount that the rule represents. |
| Target Sales Value | The average weekly total sales retail amount across all transactions in the selected periods. |
| Target Sales Profit | The average weekly total sales profit across all transactions in the selected periods. |
| Target Sales Quantity | The average weekly total sales units across all transactions in the selected periods. |
Panel
The side panel has two sections. To display the side panel, you must first select a row in the table. The top section reports some of the most useful values that are also displayed in the table. The bottom section explains the details regarding the run you are reviewing.
If 1234816 - Tully’s French Roast Dar
Support 0.13% Confidence 0.0741 Reverse Confidence 0.0945 Lift 5.4780 Sales Amt Pct Of Total 0.45% Avg Rule Sales Amt $174.42 Avg Rule Sales Profit Amt $0.00 Avg Rule Sales Oty 70.00 Avg Total Sales Amt $38,34272 Avg Tot Sales Profit Amt $4,538.56 Avg Tot Sales Qty 11,858.00 Avg Tot Txn Cnt 9,392.00
» If 1234816 - Tully’s French Roas
Results Parameters
Minimum Support 0.02%
Minimum Confidence 0.0000 Minimum Lift 0.0001 Minimum Transactions 2.00 Maximumwa 9,999.00 Maximum Rule Size 2.00 Date From) = 11/1/2015 Date To = 1/31/2016
Hier Level SKU
4 Baseline Set
» 503 - Item_Jan1_NE
4 Comparison Set
» 501 -Item_NE_Dec15
Click View to see the comparison. the View Results pop-up displays, which can be filtered by a combination of Department Item, Class, Top Results By, Subclass, and Top Results Displayed.
The results are displayed as a pair of rows for each item, a baseline row followed by a comparison row.
Table 7-20 Compare Results
| Field | Description |
|---|---|
| If | Displays the classes or subclasses containing items that were purchased in the market baskets for a given product affinity. All market baskets that contributed to a rule will contain items from these product categories. |
| Then | Displays the class or subclass that was found to have an affinity with the IF components of the market basket rule. |
| Set | Indicates whether the row of data is for the baseline runs or the comparison runs. |
| Set Size | The number of Item/Then components in the affinity rule. For example, “If Coffee and Creamer Then Sugar” has a Set Size of 3. |
| Baskets | The number of market baskets that contain items from both the IF and THEN components of the rule. |
| Avg Sales from Affinity Items | The average sales retail amount from items belonging to the THEN class or subclass. |
| Avg Profit from Affinity Items | The average sales profit from items belonging to the THEN class or subclass. |
| Avg Sales Quantity from Affinity Items | The average sales units from items belonging to the THEN class or subclass. |
| Avg Sales from “If” Items | The average sales retail amount from items belonging to the IF classes or subclasses. |
| Avg Profit from “If” Items | The average sales profit from items belonging to the IF classes or subclasses. |
| Avg Sales Quantity from “If” Items | The average sales units from items belonging to the IF classes or subclasses. |
| Support | The percent of market baskets that contain items from the IF and THEN components of the rule. |
| Confidence | The ratio of the number of transactions where the entire rule is present, compared to all transactions containing IF items. |
| Affinity Reverse Confidence | The ratio of the number of transactions where the entire rule is present, compared to all transactions containing THEN items. |
| Lift | A statistical measure of how strong the product affinity is, as compared to random chance for the rule (higher is better). |
| % MB Sales | The percent of total sales retail amount that the rule represents. |
| Avg Sales | The average weekly sales retail amount for the IF and THEN items in the rules. |
| Avg Profit | The average weekly sales profit for the IF and THEN items in the rules. |
| Avg Quantity | The average weekly sales units for the IF and THEN items in the rules. |
| Target Sales Value | The average weekly total sales retail amount across all transactions in the selected periods. |
| Target Sales Profit | The average weekly total sales profit across all transactions in the selected periods. |
Table 7-20 (Cont.) Compare Results
| Field | Description |
|---|---|
| Target Sales Quantity | The average weekly total sales units across all transactions in the selected periods. |
| Total Basket Sales Volume | The average weekly transaction count across all transactions. |
Create Custom Run
To create a custom run, select Create Custom Run from the Action drop-down. You see the Create Custom Run pop-up with two pages - Item Selection and Results Parameters. Need more info about recommendations to create a run.
Use the Item Selection page to select the items to include in the results, including the level of the results and the hierarchy nodes to include.
Provide a meaningful name for the Run Name that can help identify the run.
The Results Level defines the level of the hierarchy that the data is created for. Lower levels can require lower thresholds on the Results Parameters page (in order to find results), while higher levels can include too many results if the settings are not increased on the Results Parameters page.
The Filter Data Level is used to define rules regarding how to filter data. For example, if the results are only required for specific departments, then select Departments as the level and then select the departments.
A Promotion value of Yes only includes results in which the If component was on a promotion. A value of No is used when there is no concern about whether a promotion was present or not.
After entering item criteria, click Add Item Selection . The bottom section of the screen is populated with the results of the selection. You can also see this information in the right panel.
ee
In this guide
- Guide: AI Foundation User Guide
- Previous: 6 Attribute Extraction
- Next: 8 Size Profiles