เขตเวลาและช่วงวันที่

9 a.m. in Tokyo is not 9 a.m. in London. Real-world timestamps need a time zone so they mean the same moment everywhere. 🌍

A timestamp with no zone is called naive. One that knows its zone is aware. Mixing the two leads to silent, painful bugs.

Use tz_localize to stamp a naive time with its true zone. This says where the clock reading was actually taken.

Once aware, switch zones with tz_convert. The underlying instant stays fixed; only the displayed clock changes.

The golden order is localize first, convert second. tz_localize sets the origin zone; tz_convert moves it. Never convert a naive value.

Store data in UTC and convert only for display. This avoids confusion and sidesteps daylight-saving headaches entirely.

Need a clean sequence of dates? date_range builds one from a start, an end, and a frequency in a single call.

The freq argument controls spacing: "D" for daily, "h" for hourly, "ME" for month-end. The same rules you used to resample.

Skip the end date and pass periods to get exactly that many steps. Handy when you know the count but not the finish.

A date range makes a perfect index for a complete time series, exposing missing days you can then fill or flag.

Pair date_range with the dt accessor to manufacture calendars, fill gaps, or label every day in a forecast horizon.

You have a naive timestamp and want to show it in another zone.

You localized and converted time zones the right way, leaned on UTC as an anchor, and built clean calendars with date_range. ⏰