Dates

Date Difference

Compute the number of days between paired start and end dates. By hand, parse each ISO string to a datetime.date and subtract to get a timedelta; .days extracts the whole-day count. With pandas, subtract two datetime64 Series to get a timedelta Series, then .dt.days gives the integer day counts.

By hand

With pandas

pd.to_datetime(df['end']) - pd.to_datetime(df['start']) subtracts the two datetime64 columns, yielding a timedelta64 Series. .dt.days extracts the integer day count. The result dtype is int64 — no NaN, no float upcast.

naive.py 
import datetime
starts = ['2024-01-05', '2024-03-15', '2024-06-01', '2024-09-10']
ends   = ['2024-01-19', '2024-03-22', '2024-06-30', '2024-10-08']
result = []
for i in range(len(starts)):
    s = datetime.date.fromisoformat(starts[i])
    e = datetime.date.fromisoformat(ends[i])
    result.append((e - s).days)
print('RESULT:', result)
library.py 
import pandas as pd
from dalib.display import set_display
set_display()

starts = ['2024-01-05', '2024-03-15', '2024-06-01', '2024-09-10']
ends   = ['2024-01-19', '2024-03-22', '2024-06-30', '2024-10-08']
df = pd.DataFrame({'start': starts, 'end': ends})
d = (pd.to_datetime(df['end']) - pd.to_datetime(df['start'])).dt.days
result = d.tolist()
print('index:', d.index.tolist())
print('dtype:', d.dtype)
print('values:', d.tolist())
print('RESULT:', result)

index: [0, 1, 2, 3]
dtype: int64
values: [14, 7, 29, 28]
RESULT: [14, 7, 29, 28]

Implementation notes

  • ISO format (YYYY-MM-DD) is used throughout to avoid locale-ambiguous parsing. pd.to_datetime handles it unambiguously regardless of system locale; for non-ISO inputs use format= explicitly.
  • .dt.days gives whole days only — the fractional-day part (from non-midnight timestamps) is discarded. Use .dt.total_seconds() and divide if sub-day precision matters.
  • d.tolist() on an int64 Series returns Python int values; naive timedelta.days is also a Python int — RESULT matches without any conversion.
  • Cross-reference: diff-previous-row (python-pandas ch09) for the row-to-row lag pattern (each row vs the previous) vs the explicit paired subtraction here.