First Sign-Ins by Day

Problem

A streaming service logs each action a viewer takes, including their sign-ins, in a DataFrame `viewer_event` (`viewer_event.csv`).

```text
+-------------+--------+
| Column | Type |
+-------------+--------+
| viewer_id | int |
| action | object |
| acted_on | object |
+-------------+--------+
```

There may be many rows per viewer. The `action` column holds values such as `'signin'`, `'browse'`, and `'signout'`.

Write a function that reports, for each date within the last 90 days, how many viewers signed in for the very first time on that date. A viewer's first sign-in is the earliest date on which they have a `'signin'` action. Assume today is 2023-06-30, so only first sign-ins on or after 2023-04-01 should appear. Return columns `first_signin_date` and `new_viewers` in any order. Only include dates with at least one first-time sign-in.

Input data

Example rows — the live problem includes the full dataset.

Expected output

Your answer should return 2 rows with the columns first_signin_date, new_viewers.

Starter code (Pandas (Python))

import pandas as pd

def first_signins(viewer_event: pd.DataFrame) -> pd.DataFrame:
    # Your code here
    return viewer_event

Related Pandas questions

• 7-Day Rolling Signup AverageMedium
• Accounts Streaming From Two Devices at OnceMedium
• Agent Resolution RateMedium
• Agents With a Dominant Contact ChannelMedium
• Airbnb Popularity Bucket PricingMedium
• Amazon Unique Shipments per MonthMedium