0

I am using Pandas to read a CSV file, Forex to convert the currency to other currencies and the integer mode (int) to remove the decimal division, but it gave an error.

Sample CSV:

Item,Price (BRL)
Dining devices,100
Dishwasher,600
Electric shower,200
Fridge,1600
Induction cooktop cooker,1800
Kitchen cabinet,900
Kit pans,200
Microwave,700

And:

import pandas as pd
from forex_python.converter import CurrencyRates
from pandas.io.parsers import read_csv

cc = CurrencyRates()

cad = cc.convert('BRL', 'CAD', 1)
nzd = cc.convert('BRL', 'NZD', 1)
usd = cc.convert('BRL', 'USD', 1)

c = read_csv('data/purchases.csv')
c.loc["Total"] = c.sum()
c["Item"].values[-1] = "  "

I replaced round with int, as suggested from Python: Remove division decimal:

c["USD"] = int((((c["Price (BRL)"] * usd) / 2) * 2 + 1))
c["CAD"] = int((((c["Price (BRL)"] * cad) / 2) * 2 + 1))
c["NZD"] = int((((c["Price (BRL)"] * nzd) / 2) * 2 + 1))
c

It gave an error:

TypeError: cannot convert the series to <class 'int'>
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-4-c0af80ffd537> in <module>
     14 c["Item"].values[-1] = "  "
     15 
---> 16 c["USD"] = int((((c["Price (BRL)"] * usd) / 2) * 2 + 1))
     17 c["CAD"] = int((((c["Price (BRL)"] * cad) / 2) * 2 + 1))
     18 c["NZD"] = int((((c["Price (BRL)"] * nzd) / 2) * 2 + 1))

~/GitLab/Gustavo/global/.env/lib/python3.9/site-packages/pandas/core/series.py in wrapper(self)
    139         if len(self) == 1:
    140             return converter(self.iloc[0])
--> 141         raise TypeError(f"cannot convert the series to {converter}")
    142 
    143     wrapper.__name__ = f"__{converter.__name__}__"

TypeError: cannot convert the series to <class 'int'>
Oo'-
  • 203
  • 6
  • 22
  • You can't convert an entire series to a single `int`. Instead, your have to `apply` or `map` int to the elements of the series. With that hint, I expect you can fix the code. – Prune Jun 10 '21 at 16:21
  • @Prune *have to* is very strong i think; no need to `apply` or `map`, answer is below with `astype(int)`. But if you meant something else, pardon me. – Mustafa Aydın Jun 10 '21 at 16:26
  • Good point; yes, that's a better solution. "You can" is phrase more appropriate than "you have to". – Prune Jun 10 '21 at 16:37

1 Answers1

2

While most operations on a series are vectorized, i.e. pd.Series([n for n in ...]) + 1 means pd.Series([n + 1 for n in ...]), that is not the case of int(), which attemps to convert the full pandas.Series object to an integer. That doesn’t work.

Instead you want a pandas way of casting each element to int, try astype() for example

>>> df['Price (BRL)'] * usd
0     20.0
1    120.0
2     40.0
3    320.0
4    360.0
5    180.0
6     40.0
7    140.0
Name: Price (BRL), dtype: float64
>>> (df['Price (BRL)'] * usd).astype(int)
0     20
1    120
2     40
3    320
4    360
5    180
6     40
7    140
Name: Price (BRL), dtype: int64

I suppose your multiplication/division by 2 and adding 1 is in order to round to nearest. Casting directly to int does indeed round down. Instead you can use pd.Series.round():

>>> pd.Series([.6]).astype(int)
0    0
dtype: int64
>>> pd.Series([.6]).round().astype(int)
0    1
dtype: int64

So probably what you’re trying to achieve is (df['Price (BRL)'] * usd).round().astype(int)

Cimbali
  • 11,012
  • 1
  • 39
  • 68