during filtering of a pandas dataframe: ValueError: Buffer has wrong number of dimensions (expected 1, got 2)

Question

I can't get a hold of what is going wrong here.

d = {'x' : [1,4,6,9],
     'y' : [1,4,6,8]}
df = pd.DataFrame(d)

#filter columns based on value in specific row
df_VIP = df.iloc[:,df.iloc[1:2,:]<3]

I get the error. An this also happens with my real dataframe...

ValueError: Buffer has wrong number of dimensions (expected 1, got 2)

I hope anyone has a hint to solve it. Thanks!

Answer 1

If possible, select by one row, e.g. second by 1 with convert to numpy array, because used iloc:

d = {'x' : [1,4,6,9],
     'y' : [1,2,6,8]}
df = pd.DataFrame(d)

df_VIP = df.iloc[:,df.iloc[1,:].to_numpy()<3]
print (df_VIP)
   y
0  1
1  2
2  6
3  8

If use your select - one row DataFrame instead Series flatten values by indexing:

print (df.iloc[1:2,:])
   x  y
1  4  2


print (df.iloc[1:2,:].to_numpy())
[[4 2]]

print (df.iloc[1:2,:].to_numpy()[0])
[4 2]

df_VIP = df.iloc[:,df.iloc[1:2,:].to_numpy()[0]<3]