Uncovering the associations of different products can help optimize stocks, store arrangements or product recommendations. One can quickly discover these associations using Actable AI's Association Rules within minutes.
In this example, we shall look at this data set which contains records of grocery purchases. Each record in the data set contains a buyer's id and a bought item. Using association rules to look at items bought by the same buyers, we shall uncover which items are usually bought together.
After uploading the data set and selecting Asssociation Rules, we have to select a column or a list of columns to group items that are bought together. In this example, it is Member Number which indicates the buyer who purchased the items. There are also other options to adjust for item popularity and filter items by different metrics. After selecting these options and run the analysis, the results will be shown in different charts.
The association graph shows the associations and their strength between different items. The thickness of the edges represents the confidence of the associations. Here we can easily see that ones buying root vegetables likely also buy whole milk, or ones buying yogurt will likely buy rolls/buns. It is worth noting that the associations are not symmetric, for example, ones buying whole milk unlikely aslo buy root vegetables.
Chord digram shows the associations with items arranged radially around a cirle. The arcs connecting two items show association strengths between two items. Note that the widths of arc ends are different as assoications are asymmetric.
The animation below shows an example of how our Assocation Rules works with the above data set.
To learn more about how our Association Rules analysis works, please check out our documentation here.
