Error in Loop: "Cannot unpack non-iterable Timedelta object"-CodePudding

I'm in the process of learning Python, and I'm trying to make a simple loop, for adding dirty prices, to my dataframe bond_df.

Days_left is a Series, bond_df is a pandas dataframe containing the closing prices used in the formula below.

If i run the command:

days = days_left[1].days

I get an integer of size 1 with the value of 2, and this is exactly what I need. I need the value of days as integers, and without any other time-stamp on it (see the attached picture). So, I use ".days", so that I can extract the integer value of the days, and get rid of the time-stamp for hours and seconds etc.

Because of this, i figured I could then use this in a loop to construct my column of dirty-prices, in my df:

for i, number in days_left:
    days = days_left[i].days
    bond_df['dirty_price'][i] = bond_df['closing_price'][i]   ((365 - days)/365)

However this does not work and returns the message:

"TypeError: cannot unpack non-iterable Timedelta object"

I then figured, that I could construct a loop using a range instead:

for i in range(0, len(days_left)):
    days = days_left[i].days
    bond_df['dirty_price'][i] = bond_df['closing_price'][i]   ((365 - days)/365)
    print(days, bond_df['dirty_price'])

This seem to work as intended.

But I would still like to find out, what I did wrong in the first instance.

Can somebody explain the difference between these two loops and why I cannot do as above?

All the best, Nic

CodePudding user response：

you can simplify by using pandas vectorized functionality:

import pandas as pd

DAYS_IN_YEAR = 365 # this actually isn't constant; adjust as needed

df = pd.DataFrame(
    {
        "days_left": [pd.Timedelta(days=1), pd.Timedelta(days=2), pd.Timedelta(days=3)],
        "closing_price": [1, 2, 3],
    }
)

df["dirty_price"] = df["closing_price"]   (
    (DAYS_IN_YEAR - df["days_left"].dt.total_seconds() / 86400)
    / DAYS_IN_YEAR
    # could also use df["days_left"].dt.days here if hours minutes etc. don't matter
)

df
  days_left  closing_price  dirty_price
0    1 days              1     1.997260
1    2 days              2     2.994521
2    3 days              3     3.991781

CodePudding user response：

Here:

for i, number in days_left:

you meant

for i, number in enumerate(days_left):

most likely...

CodePudding user response：

What you have is likely something like this:

In [20]: dt = pd.timedelta_range("1 day", "5 day", freq="1d")
In [21]: type(dt)
Out[21]: pandas.core.indexes.timedeltas.TimedeltaIndex

In Python, iteration gives you the values:

In [22]: [i for i in dt]
Out[22]: 
[Timedelta('1 days 00:00:00'),
 Timedelta('2 days 00:00:00'),
 Timedelta('3 days 00:00:00'),
 Timedelta('4 days 00:00:00'),
 Timedelta('5 days 00:00:00')]

To also access the index during iteration, you need to use enumerate.

In [23]: [(i, day) for i, day in enumerate(dt)]
Out[23]: 
[(0, Timedelta('1 days 00:00:00')),
 (1, Timedelta('2 days 00:00:00')),
 (2, Timedelta('3 days 00:00:00')),
 (3, Timedelta('4 days 00:00:00')),
 (4, Timedelta('5 days 00:00:00'))]

That said, you should not be iterating. Use vector operations; something like this:

bond_df['dirty_price'] = bond_df['closing_price']   ((365 - days_left)/365)