Build a monthly cohort retention matrix

Question

Growth wants classic cohort retention: users grouped by their signup month, then counted by how many months later they were still active.

Write `retention(df)` that returns a DataFrame indexed by signup month period, with integer month offsets as columns and the count of distinct active users in ea

Accepted Answer

Step 1: Subtracting two `Period` values gives an offset object; `.n` pulls out the integer number of months, which is the column axis of every cohort chart. Step 2: `aggfunc='nunique'` counts distinct users — a user with three visits in month 2 must still count once. Step 3: `fill_value=0` renders 'nobody came back' as a zero rather than a hole in the triangle. Step 4: Sorting both axes gives the familiar upper-triangular shape where each row starts at offset 0.

Build a monthly cohort retention matrix

More Pandas — GroupBy questions

Asked at Swiggy