pandas/python: creating a numerical categorical variable that counts the categories

Question

I am trying to build a column in a pandas DF that is counting the category CHANGES of a categorical variable in a "rolling" way. What I keep on finding in stackoverflow is a number of rolling counts, which is exactly the opposite of what I am looking for. I am looking for a column that runs through an alphabetically sorted categorical column and adds an increment of 1 every time the category changes but gets dragged unchanged otherwise. So if I have the variable named 'cat_var' in the example below, I need to programmatically create the column 'category_counter_var' which I manually created in the example below. Can someone help?

import pandas as pd

df = pd.DataFrame({'cat_var':['Q1','Q1','Q1','Q2','Q2','Q3','Q4','Q4','Q4','Q4']
                   ,'category_counter_var':[1,1,1,2,2,3,4,4,4,4]})

Answer 1

Use:

df['new'] = df['cat_var'].ne(df['cat_var'].shift()).cumsum()
print(df)

# Output
  cat_var  category_counter_var  new
0      Q1                     1    1
1      Q1                     1    1
2      Q1                     1    1
3      Q2                     2    2
4      Q2                     2    2
5      Q3                     3    3
6      Q4                     4    4
7      Q4                     4    4
8      Q4                     4    4
9      Q4                     4    4