The Find overlap step finds matching or non-matching rows between two tables.
This step takes two tables as its input and shows you how they overlap. It will help you figure out questions like, "Which rows from Table 1 exist in Table 2?" or "Which rows from Table 1 don't exist in Table 2?"
Table 1 shows products in our first store with three columns: "Product ID", "Name", and "Sales Price".
Table 2 shows products in our second store with two columns: "Product ID" and "Name".
Let's say we want to find which products from Store 1 that are also sold in Store 2. The step will immediately tell us this as its output. Below we can see that there are three products sold in both stores. Since "Store 1" was our primary table in the step, it's showing us the complete rows from that table.
When you first connect your two inputs into this step, you'll see that your first input is treated as the primary table. By default, the step will consider a match when any of the rules are true.
First, make sure the step is configured to use the correct primary table. Then, determine how the rules should be combined. If you are setting multiple rules, they can be combined so that any rule triggers a match or all rules must be met to be considered a match.
Now we're ready to configure the rules for how the step should identify matches. Each rule looks at one column from per table and compares them. First, select the column from each table to be compared. Then, determine if we're looking for rows that match or do not match. Finally, choose one of three options in the match type:
Exactly will look for an exact match, but will ignore casing (upper case v. lower case). Approximately will look for a fuzzy match. Case-sensitively will look for an exact match that includes casing.
If you choose approximately, you can adjust the confidence percentage. The higher the percentage, the closer to an exact match this step will attempt to make. 100% confidence will act the same as an exact match.
To add another rule, click on the + Add Rule link below the first one.