In PostgreSQL the lag() function returns the value from a previous row, specified by the given offset.
The offset specifies how many rows prior to the current row to get the value from. For example, an offset of 1 gets the value from the previous row.
In PostgreSQL the nth_value() function is a window function that returns the value from the given row of the current window frame. We provide the column and row number as an argument when we call the function.
In PostgreSQL, we can use the cume_dist() function to return the cumulative distribution of a value within a group of values.
It calculates this as follows: (the number of partition rows preceding or peers with current row) / (total partition rows).
The return value ranges from 1/N to 1.
In PostgreSQL, we can use the percent_rank() function to return the relative rank of each row, expressed as a percentage ranging from 0 to 1 inclusive.
In PostgreSQL, the ntile() function is a window function that divides a partition into the specified number of groups (buckets), distributing the rows as equally as possible, and returns the bucket number of the current row within its partition.
While it’s true that we can use the SQL MAX() function to get the maximum value in a column, what if we want to return the value from another column instead of the actual maximum value itself? And what if we want it partitioned by category, so that it’s based on the maximum value from each category? And what if we want all values listed out, including those that aren’t the most populous?
In the following example, we use a window function to solve this problem.
Ever since the release of MySQL 8.0.16, we’ve had the ability to create CHECK constraints in MySQL. At some point, we may want to return a list of CHECK constraints that have been created in a database or against a given table.
Fortunately, we can use the information_schema.check_constraints view to do just that.
We can alternatively use the information_schema.table_constraints view to get the same info.
PostgreSQL has a window function called rank() that returns the rank of the current row, with gaps.
“With gaps” means that it returns the same rank for any ties (i.e. two or more rows with the same value), but then subsequent ranks jump forward to account for the ties.
This means that there’s the potential for noncontiguous rank values. For example it could go 1, 2, 5, etc if several rows are ranked at 2. If there are no ties, then the rank values will be contiguous.
In PostgreSQL, we can use the row_number() function to get each row’s number within its partition. This allows us to create a column with incrementing row numbers that reset with each new partition.
The row_number() function is a window function that’s specifically designed to return the number of the current row within its partition, starting at 1 and incrementing sequentially.