CODEWITHSUNDEEP
pythonpandasnumpydataframe
Helix Herry
Helix Herry

Reputation: 327

TypeError: must be str, not int couldn't be solved

The original time data is like this：

df['time'][0:4]

   2015-07-08
        05-11
        05-12
   2008-07-26

I want all these data contains year value. And I applied this:

con_time = []

i=0
for i in df['time']:
    if len(df['time'])==5:
        time = '2018'+'-'+df['time']
        con_time.append(time)
        i +=1
    else:
        con_time.append(df['time']) 
        i +=1

Error occurred:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-78-b7d87c72f412> in <module>()
      8     else:
      9         con_time.append(df['time'])
---> 10         i +=1

TypeError: must be str, not int

This error is so strange.... Actually I want to create a new list, converting it to a np.array and concat it into the df. Do I have a better way to achieve the goal?

Upvotes: 2

Views: 1148

Answers (2)

Mischa Lisovyi
Mischa Lisovyi

Reputation: 3223

Since you have asked about an alternative approach. instead of an explicit loop in python and filling a list, one should rather use DataFrame methods directly. In your case this would be

df['time'].apply(lambda x: x if len(x) != 5 else '2018-'+x)

This might run faster for some datasets

EDIT I actually ran a timing benchmark using a random toy dataset with ~50% of complete and incomplete dates. In short, it seems that for a small dataset the simple for-loop solution is faster for a large dataset both methods show similar performance:

# 1M examples
import random
import numpy as np
y = pd.Series(np.random.randint(0,2,1000000))
s = {0:'2015-07-08',  1:'05-11'}
y = y.map(s)
%%timeit -n100
_ = y.apply(lambda x: x if len(x) != 5 else '2018-'+x)
>>> 275 ms ± 6.42 ms per loop (mean ± std. dev. of 7 runs, 100 loops each)
%%timeit -n100
con_time = []
for i in y:
    if len(i)==5:
        time = '2018-'+i
        con_time.append(time)
    else:
        con_time.append(i) 
con_time_a = np.array(con_time)
>>> 289 ms ± 5.23 ms per loop (mean ± std. dev. of 7 runs, 100 loops each)

# 1K examples
import random
import numpy as np
y = pd.Series(np.random.randint(0,2,1000))
s = {0:'2015-07-08',  1:'05-11'}
y = y.map(s)
%%timeit -n100
_ = y.apply(lambda x: x if len(x) != 5 else '2018-'+x)
>>> 431 µs ± 70.3 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)
%%timeit -n100
con_time = []
for i in y:
    if len(i)==5:
        time = '2018-'+i
        con_time.append(time)
    else:
        con_time.append(i) 
con_time_a = np.array(con_time)
>>> 289 µs ± 40.4 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

Upvotes: 3

Druta Ruslan
Druta Ruslan

Reputation: 7412

you have two i variables, when you make i += 1 you take i variable from 

for i in df['time']

not 

i = 0

change i variable from for loop with another name , for example if you don't need variable from for loop statment you can name it _ (underscore)

Upvotes: 2

Related Questions