This lesson's interactive features are locked, please to keep using them

Cleaning Dates

Dates are messy because everyone writes them differently. 2024-01-15, 01/15/2024, and Jan 15, 2024 are all the same day, but completely different strings.

The pd.to_datetime() function can convert dates into a single datetime64[ns] type. Pass format="mixed" when the input contains multiple formats. It can also handle invalid dates with the errors="coerce" option, which turns them into NaT ("Not a Time").

# Parse mixed formats
df["date"] = pd.to_datetime(df["date"], format="mixed")

# Coerce invalid values to NaT
df["date"] = pd.to_datetime(df["date"], format="mixed", errors="coerce")

Assignment

Complete the parse_device_timestamps function. It accepts a DataFrame with a timestamp (string) column, then returns a DataFrame with parsed datetimes and invalid timestamp rows removed.

Create a copy of the DataFrame to avoid modifying the original.
Convert the mixed-format timestamp column to a datetime type with format="mixed", coercing invalid values to NaT.
Drop rows with invalid dates. A reading with no usable timestamp can't be placed in the timeline.
Return the cleaned DataFrame.