How to deal with this logic in pandas

问题

I have a data frame like following below.

   coutry      flag
0  China       red
1  Russia      green
2  China       yellow
3  Britain     yellow
4  Russia      green
......................

In df['country'], you can see many different country names. I want to set the first appear country as 1, the second as 2. The flag is the same logic.So you can see the result is:

 coutry      flag
0  1          1
1  2          2
2  1          3
3  3          3
4  2          2

But I don't know how to achieve this logic in python. Thank you. Moreover when I get the result data frame, I want to have an function to back data frame to the original.

回答1:

You can use factorize and add 1:

df['coutry'] = pd.factorize(df.coutry)[0] + 1
df['flag'] = pd.factorize(df.flag)[0] + 1
print (df)
   coutry  flag
0       1     1
1       2     2
2       1     3
3       3     3
4       2     2

Then you can convert columns to categories by Categorical if need save memory:

df['coutry'] = pd.Categorical(pd.factorize(df.coutry)[0] + 1)
df['flag'] =  pd.Categorical(pd.factorize(df.flag)[0] + 1)
print (df)
  coutry flag
0      1    1
1      2    2
2      1    3
3      3    3
4      2    2
print (df.dtypes)
coutry    category
flag      category
dtype: object

#1000 times larger df
df = pd.concat([df]*1000).reset_index(drop=True)
df['coutry'] = pd.Categorical(pd.factorize(df.coutry)[0] + 1)
df['flag'] =  pd.factorize(df.flag)[0] + 1
print (df)
     coutry  flag
0         1     1
1         2     2
2         1     3
3         3     3
4         2     2
5         1     1
6         2     2
...
...

print (df['coutry'].nbytes)
5024

print (df['flag'].nbytes)
20000

If need convert back, you can map values by dictionaries:

b = [list(x) for x in pd.factorize(df.coutry.drop_duplicates())]
d1 = dict(zip(b[0], b[1]))
print (d1)
{0: 'China', 1: 'Russia', 2: 'Britain'}

b = [list(x) for x in pd.factorize(df.flag.drop_duplicates())]
d2 = dict(zip(b[0], b[1]))
print (d2)
{0: 'red', 1: 'green', 2: 'yellow'}


df['coutry'] =  pd.Categorical(pd.factorize(df.coutry)[0])
df['flag'] =  pd.Categorical(pd.factorize(df.flag)[0])
print (df)
   coutry  flag
0       0     0
1       1     1
2       0     2
3       2     2
4       1     1

df['coutry'] = df.coutry.map(d1)
df['flag'] = df.flag.map(d2)
print (df)
    coutry    flag
0    China     red
1   Russia   green
2    China  yellow
3  Britain  yellow
4   Russia   green

来源：https://stackoverflow.com/questions/38966912/how-to-deal-with-this-logic-in-pandas

标签

python

pandas

dataframe

categorical-data