Groupby column and based on that groupby another

userU1lC 0
Viewed 32

I have DataFrame like this

d = {'id': [1, 2, 3, 4, 5, 6],
     'y_true': [0, 0, 1, 1, 1, 0],
     'y_pred': [0.23, 0.01, 0.19, 0.01, 0.3, 0.23]
    }
df = pd.DataFrame(data=d)

enter image description here

And I want to groupby y_pred and then groupby y_true for the same columns to find mean y_true for each row, corresponding to y_pred. So to speak

d1 = {'y_true': [0, 0.5, 1, 1],
     'y_pred': [0.23, 0.01, 0.19, 0.3]
    }
df1 = pd.DataFrame(data=d1)

enter image description here

I know how groupby y_pred column, but I can only groupby y_true manually, row-by-row

group-by dataframe pandas python
1 Answers

try:

df.groupby('y_pred')['y_true'].mean().reset_index()
# df.groupby("y_pred").apply(lambda x: x['y_true'].mean()).reset_index(name="y_true") #same

    y_pred  y_true
0   0.01    0.5
1   0.19    1.0
2   0.23    0.0
3   0.30    1.0

#or use numpy mean (maybe numpy has higher probability to be less wrong than panda mean)
import numpy as np
df.groupby('y_pred').agg({'y_true': np.mean}).reset_index()  

#can combine both numpy mean and pandas mean
df.groupby('y_pred').agg(y_true_pd_mean=('y_true', 'mean'), y_true_np_mean=('y_true', np.mean)).reset_index()

    y_pred  y_true_pd_mean  y_true_np_mean
0   0.01    0.5             0.5
1   0.19    1.0             1.0
2   0.23    0.0             0.0
3   0.30    1.0             1.0

#can also use mean from statistics module:
import statistics
df.groupby('y_pred').agg({'y_true': statistics.mean}).reset_index()
khaled koubaa0
Related