How to convert the continuous numbers into categorical using pandas?

user_5037278 0
Viewed 490

I am searching for solution for this question for few days but unfortunately no success.

I have continuous values in a column like this:

Val: 1, 15, 2, 91, 52, 126

I need to convert these numbers into categories as intervals. For example, the first number should lie in category (1-10)

I know we can define interval and convert the data using pd.cut

pd.cut(df.val, right=False)

but my problem is that I can't define interval as I have millions of value.

The ideal solution would be that I can define the range of interval and then it automatically search the values and convert it in that category.

This would be my ideal output:

Val     Val_Cat
1        1-10
15       10-20
2        1-10
91       90-100
52       50-60
126      120-130
pandas categorical-data cut python
2 Answers

You can create a new column with the desired respective category:

def cat(x): 
    return str(x//10*10) + '-' + str((x//10*10)+10)

df['Val_cat']=list(map(cat, df.Val))
user_82285580

One idea is use maths with integer division by // by 10, then multiple by 10 and last convert to strings (with repalce if necessary):

s = df['Val'] // 10 * 10
df['new'] = s.replace(0, 1).astype(str) + '-' + (s + 10).astype(str)
print (df)
   Val  Val_Cat      new
0    1     1-10     1-10
1   15    10-20    10-20
2    2     1-10     1-10
3   91   90-100   90-100
4   52    50-60    50-60
5  126  120-130  120-130

Alternative with f-strings:

df['new'] = df['Val'].map(lambda x: f'{x//10*10}-{(x//10*10)+10}')
print (df)
   Val  Val_Cat      new
0    1     1-10     0-10
1   15    10-20    10-20
2    2     1-10     0-10
3   91   90-100   90-100
4   52    50-60    50-60
5  126  120-130  120-130

Your solution with cut is possible change by:

bins = np.arange(0, df['Val'].max() // 10 * 10 + 20, 10)

df['new'] = pd.cut(df.Val, bins = bins, right=False)
print (df)
   Val  Val_Cat         new
0    1     1-10     [0, 10)
1   15    10-20    [10, 20)
2    2     1-10     [0, 10)
3   91   90-100   [90, 100)
4   52    50-60    [50, 60)
5  126  120-130  [120, 130)
user_29010020
Related