I have a dataframe like:
df
name value
0 a 10
1 a 5
2 b 4
3 b 8
I want to create filtered datasets using the 'name' feature. Something like:
for i in ['a', 'b']:
df_{i} = df[df['name'] == i]
df_a
name value
0 a 10
1 a 5
df_b
name value
2 b 4
3 b 8
Which alterations are necessary to do that?
You can use unique() and save those in a dict() (fdf):
import pandas as pd
data = {
'name': ['a', 'a', 'b', 'b'],
'value': [10, 5, 4, 8]
}
df = pd.DataFrame(data)
fdf = {}
for i in df['name'].unique():
fdf[f'df_{i}'] = df[df['name'] == i]
df_a, df_b = fdf['df_a'], fdf['df_b']
print("df_a:")
print(df_a)
print("\ndf_b:")
print(df_b)
df_a:
name value
0 a 10
1 a 5
df_b:
name value
2 b 4
3 b 8
Code
Creating variables dynamically is not recommended in Python.
Create dict for that purpose.
dictionary comprehension
dict_grouped = {k: d for k, d in df.groupby('name')}
dict_grouped:
{'a': name value
0 a 10
1 a 5,
'b': name value
2 b 4
3 b 8}
dict func
Alternatively, you can also use the following code instead of dictionary comprehension.
dict_grouped = dict(tuple(df.groupby('name')))
If you absolutely must create it dynamically as a variable, you can use the code below, but i don't recommend it.
for k, d in df.groupby('name'):
globals()[f'df_{k}'] = d
df_a:
name value
0 a 10
1 a 5
