Converting NumPy array to a column vector

Question

Subsetting a column of a DataFrame gives me the y (dependent) variable in form of a NumPy array.

y = train['Survived']

But printing the .shape of the variable y (y.shape) outputs (891,) (notice it's not (891, 1), a column vector).

I would like to perform matrix multiplication of y with a variable with size (1 x 10) using np.matmul, but it's throwing me this error:

Exception: Dot product shape mismatch, (891,) vs (1, 10)

How can I force the y variable to be a column vector with size (891, 1) instead of just (891, )?

y = train[['Survived']] will also have that column vector shape. However that is quite a bit slower, since it makes a new dataframe (as opposed to extracting a series/column). — hpaulj
– hpaulj, Commented Nov 17, 2020 at 5:53

Mikhail Genkin · Accepted Answer · 2020-11-17 05:37:27Z

3

Just use y[:,None]. This will have the correct shape

answered Nov 17, 2020 at 5:37

Mikhail Genkin

3,4704 gold badges29 silver badges55 bronze badges

Sign up to request clarification or add additional context in comments.

1 Answer 1