Hey there, let's tackle this time difference issue you're facing in SPSS Modeler. I've run into similar quirks with the TIME_HOURS_DIFFERENCE function before, so here are a few targeted checks and fixes to try:

1. Double-check the parameter order of TIME_HOURS_DIFFERENCE #

This function follows the syntax TIME_HOURS_DIFFERENCE(time1, time2), which calculates time1 minus time2 in hours. If you want the positive hour difference between arrival and departure (arrive time minus depart time), make sure your function is written as:

TIME_HOURS_DIFFERENCE(arrive_Date, Depart_Date)

If you accidentally reversed the parameters (e.g., TIME_HOURS_DIFFERENCE(Depart_Date, arrive_Date)), cross-day calculations will naturally return negative values since departure time is earlier than arrival time.

2. Ensure your fields are full Date/Time types (not just Time) #

If arrive_Date and Depart_Date only store time values (no date component), even with "Rollover days/mins" enabled, Modeler can't recognize cross-day logic. For example, if arrival time is 08:00 (next day) but stored as a standalone time field, the system treats it as the same day's 08:00—subtracting the previous day's 20:00 (also stored as same-day time) will give -12 instead of the correct positive value.

• Fix: Verify both fields are set to the Date/Time type (with full year-month-day and time components). If you have separate date and time fields, merge them using:

  DATE_AND_TIME(arrive_Date_field, arrive_Time_field)
  

3. Confirm the "Rollover days/mins" setting is properly applied #

Even though you mentioned enabling this, it's worth a quick recheck:

• Go to Edit > Stream Properties > Data tab
• Ensure the box for "Rollover days/mins for time calculations" is checked
• After adjusting this setting, re-run your entire stream (not just the calculation node) to make sure the change takes effect.

4. Alternative calculation using DATE_DIFF (if all else fails) #

If the function still behaves unexpectedly, you can explicitly combine date and time calculations to bypass rollover ambiguities:

DATE_DIFF(arrive_Date, Depart_Date, "days") * 24 + TIME_HOURS_DIFFERENCE(arrive_Date, Depart_Date)

This formula first calculates the full day difference (converted to hours) and adds the remaining time-hour difference. For example:

• Arrive: 2024-05-20 08:00, Depart: 2024-05-19 20:00
• DATE_DIFF returns 1 day → 1*24 = 24 hours
• TIME_HOURS_DIFFERENCE returns -12 hours (08:00 - 20:00)
• Total: 24 + (-12) = 12 hours (correct positive value)

内容的提问来源于stack exchange，提问作者Saud