Python Pandas: TypeError: unsupported operand type(s) for +: 'datetime.time' and 'Timedelta'

Question

I want to sum two pandas Series. The first series is a hh:mm time value; the second series contains Timedelta objects obtained from pd.Timedelta.

The expected result is a hh:mm time regardless of day change (e.g. 22:00 + 4 hours = 02:00).

I created the Delta series like this:

delta = pd.Series(0 for x in range(0, len(df.Time_In_Hours)))

for j in range(0, len(df.Time_In_Hours)):
    delta[j] = pd.Timedelta(df.Time_In_Hours[j], 'h')
df = df.assign(Delta = delta)   
print ("Delta dtype = %s" % (df.Delta.dtype))
print ("Start_Time dtype = %s" % (df.Start_Time.dtype))

#Output
Delta dtype = object
Start_Time dtype = object

What I tried:

df["end_Time"] = df["Start_Time"] + df["Delta"]  

The error I am receiving is: TypeError: unsupported operand type(s) for +: 'datetime.time' and 'Timedelta'

How can I sum these two series ?

Answer 1

The cause

The error is pretty clear. If you check the types of the elements, you will find out that at some point you are tying to add datetime.time object and pandas.Timedelta.

There are 2 kinds of dates, times and timedeltas:

• python's builtin from datetime module i.e. datetime.time, datetime.date, datetime.timedelta, ...
• pandas / numpy i.e pandas.Timestamp, pandas.Timedelta

these two stacks are incompatible for basic operations as addition or comparison.

Solution 1

Convert everything to pandas type and extract the times in the end

You should make sure, that dtypes of your columns are something like datetime64[ns] and timedelta64[ns]. For that, try converting them explicitly using pd.to_datetime and pd.to_timedelta.

Solution 2

Another approach would be just converting the Delta column to datetime.timedelta you could try

df["end_Time"] = df["Start_Time"] + df["Delta"].map(pd.Timedelta.to_pytimedelta)

But you may run into some more errors depending on what is in your df["Delta"] and df["Start_Time"]