I have a column col a that can have values listed multiple times.
For each distinct value in col_a I would like to generate a corresponding random value in col_b. Like so:
col_a col_b
A 0.25
A 0.25
B 0.12
B 0.12
How can I generate col_b?
You can call random.random() for each group:
import random
df.groupby('col_a')['col_a'].transform(lambda x: random.random())
Out:
0 0.394776
1 0.394776
2 0.928343
3 0.928343
Name: col_a, dtype: float64
Assign it back:
df['col_b'] = df.groupby('col_a')['col_a'].transform(lambda x: random.random())
df
Out:
col_a col_b
0 A 0.012639
1 A 0.012639
2 B 0.839752
3 B 0.839752
I'd do it this way:
import pandas as pd
import random
df['col_b'] = 1
df['col_b'] = df.groupby('col_a')['col_b'].transform(lambda _:random.random())
Create numbers for the number of unique values in col_a and index it with a factorized version of col_a
u, f = np.unique(df.col_a.values, return_inverse=True)
df.assign(col_b=np.random.rand(u.size)[f])
col_a col_b
0 A 0.470264
1 A 0.470264
2 B 0.836461
3 B 0.836461
For large data, this is quicker
f, u = pd.factorize(df.col_a.values)
df.assign(col_b=np.random.rand(u.size)[f])
col_a col_b
0 A 0.476353
1 A 0.476353
2 B 0.639068
3 B 0.639068
