Set values in dataframe with multiple element-wise conditions

Question

I have a dataframe. I want to set all values that are less than x, in rows that have a certain parameter value, to a new value.

I try this, but nothing happens to my dataframe.

   df_data = {'A': [4, 4, 5, 5],
               'B': [4, 4, 4, 5],
               'C': [4, 5, 5, 5],
               'Bool': [True, True, False, False]}
    test_df = pd.DataFrame(df_data, columns=['A', 'B', 'C', 'Bool'])
    test_df[test_df.iloc[:, :-1] < 5][test_df['Bool'] == True] = 99
    print(test_df)

I expect some elements in the test_df to have value 99.

You are absolutely right, because I forget `loc`, sorry - `test_df.loc[(test_df.iloc[:, :-1] < 5) & test_df['Bool'], 'A'] = 99` — jezrael, Sep 03 '19 at 12:51
I get error message: ValueError: operands could not be broadcast together with shapes (16,) (4,) — muzzex, Sep 03 '19 at 12:55
yop, I see it. what is expcted output from your sample data? — jezrael, Sep 03 '19 at 12:56
df_data = {'A': [99, 99, 5, 5], 'B': [99, 99, 99, 5], 'C': [4, 5, 5, 5], 'Bool': [True, True, False, False]} — muzzex, Sep 03 '19 at 12:56

jezrael · Accepted Answer · 2019-09-04T06:04:59.833

1

If want chain boolean DataFrame with & for bitwise AND is necessary convert it to numpy array (N, 1) and then set new values with DataFrame.mask:

m = (test_df.iloc[:, :-1] < 5).values & test_df['Bool'].values[:, None]
print (m)
[[ True  True  True]
 [ True  True False]
 [False False False]
 [False False False]]

test_df.iloc[:, :-1] = test_df.iloc[:, :-1].mask(m, 99)
print(test_df)
    A   B   C   Bool
0  99  99  99   True
1  99  99   5   True
2   5   4   5  False
3   5   5   5  False

edited Sep 04 '19 at 06:04

answered Sep 03 '19 at 13:09

jezrael

822,522
95
1,334
1,252

I get error message when creating m: ValueError: Unable to coerce to DataFrame, shape must be (4, 3): given (4L, 1L) – muzzex Sep 03 '19 at 14:53
m = (test_df.iloc[:, :-1] < 5) & (pd.DataFrame(pd.np.tile((test_df['Bool'].values[:, None] == True), (1, 3)))).values test_df.iloc[:, :-1] = test_df.iloc[:, :-1].mask(m, 99) – muzzex Sep 03 '19 at 15:05
@muzzex can you test `m = (test_df.iloc[:, :-1].values < 5) & test_df['Bool'].values[:, None] print (m)` – jezrael Sep 03 '19 at 15:35
1

That last comment also work. Need the ".values". Thanks jezrael for your help! – muzzex Sep 04 '19 at 05:58

score 0 · Answer 2 · answered Sep 03 '19 at 15:07

0

This works:

m = (test_df.iloc[:, :-1] < 5) & (pd.DataFrame(pd.np.tile((test_df['Bool'].values[:, None] == True), (1, 3)))).values
test_df.iloc[:, :-1] = test_df.iloc[:, :-1].mask(m, 99)

answered Sep 03 '19 at 15:07

muzzex

157
2
16

Set values in dataframe with multiple element-wise conditions

2 Answers2