Converting dates into day of year in pandas data frame

user_14009369 0
Viewed 87

In the data frame df1, how to convert q1,....q9,q0,...d5 days into day of years?

YYYY,MM,q1,q2,q3,q4,q5,q6,q7,q8,q9,q0,d1,d2,d3,d4,d5
1975,01,2,11,12,26,25,10,29,21,30,22,8,7,14,4,13
1975,02,27,22,8,20,6,26,21,4,19,9,10,1,11,12,23
1975,03,8,7,21,22,25,9,4,30,2,19,10,11,28,12,27
1975,04,29,28,27,17,19,2,30,16,18,3,9,10,11,8,13

What I have tried is

from datetime import datetime
day_of_year = datetime.now().timetuple().tm_yday

But it is not working. Please help.

jupyter-notebook pandas python-2.7 math python
3 Answers

IIUC:

# Flat your dataframe to vectorize datetime operation
out = df.melt(['YYYY', 'MM'], ignore_index=False).astype(str)

# Compute day of year
out['value'] = pd.to_datetime(out['YYYY']+'-'+out['MM']+'-'+out['value']).dt.dayofyear

# Reshape your dataframe as your original
df = out.reset_index().pivot(['index', 'YYYY', 'MM'], 'variable', 'value') \
        .droplevel(0).reset_index().rename_axis(columns=None)[df.columns]

Output:

>>> df
   YYYY MM   q1   q2   q3   q4   q5  q6   q7   q8   q9  q0  d1   d2   d3  d4   d5
0  1975  1    2   11   12   26   25  10   29   21   30  22   8    7   14   4   13
1  1975  2   58   53   39   51   37  57   52   35   50  40  41   32   42  43   54
2  1975  3   67   66   80   81   84  68   63   89   61  78  69   70   87  71   86
3  1975  4  119  118  117  107  109  92  120  106  108  93  99  100  101  98  103
user_152399510

You can subtract (January 1st) of the same year from a date, to get the day of the year:

from datetime import datetime

def day_of_year(y, m, d):
    return (datetime(y, m, d) - datetime(y, 1, 1)).days + 1

print(day_of_year(1975,2,27))
# 58
user_30807230

You can avoid wide to long and back to wide by using apply; you should get some speedup by avoiding the round trip:

outcome = (df
           .filter(regex = r".+\d$")
           .astype(str)
           .apply(lambda m: pd.to_datetime(df.YYYY.astype(str) 
                                           + "-" 
                                           + df.MM.astype(str) 
                                           + "-" 
                                           + m)
                                           .dt.dayofyear)
           )

df.assign(**outcome)


   YYYY  MM   q1   q2   q3   q4   q5  q6   q7   q8   q9  q0  d1   d2   d3  d4   d5
0  1975   1    2   11   12   26   25  10   29   21   30  22   8    7   14   4   13
1  1975   2   58   53   39   51   37  57   52   35   50  40  41   32   42  43   54
2  1975   3   67   66   80   81   84  68   63   89   61  78  69   70   87  71   86
3  1975   4  119  118  117  107  109  92  120  106  108  93  99  100  101  98  103
user_71757130
Related