Vectorizing a function using column and row index

Question

I have built a program that takes values out of one dataframe, and use these as an input for another dataframe. df_coordinate is a dataframe with rows of x and y coordinates (with sizes between 0 and max_size_x or max_size_y). updated_coordinates is a new dataframe with the size of the screen, that uses the coordinates and a sort of Euclidean distance to make the coordinates represent an area instead of a pixel. The code works as I intend, but right now it is way to slow. I know that vectorizing is much faster, and I try to implement it as much as possible. However, I can't seem to find a way how to vectorize with formulas where the column and row index is used. With .apply I use x.name and x.index as you can see, but is there a way to implement this faster?

max_size_x = 1080
max_size_y = 720

    for index, row in df_coordinate.iterrows():
        updated_coordinates = pd.DataFrame(np.zeros((max_size_x, max_size_y))) # to do: maybe empty instead of zeroes, and already delete the ones from attention_max
        current_time = df_coordinate.loc[index]
        coordinate_x = current_time['x_coordinate']
        coordinate_y = current_time['y_coordinate']

        # calculate area with Euclidean distance:
        updated_coordinates = updated_coordinates.apply(lambda x: 1/ ( np.power((np.sqrt(np.power(x.name-coordinate_x,2) + np.power(x.index-coordinate_y,2))),2)))

Bram Zijlstra · Accepted Answer · 2017-08-18 10:33:11Z

1

I found the answer, by adding this outside of the loop:

df_x = pd.DataFrame(np.zeros((image_size_x, image_size_y))) 
df_x = df_x.apply(lambda x: x.name) 
df_y = pd.DataFrame(np.zeros((image_size_x, image_size_y))) 
df_y = df_y.apply(lambda x: x.index)

And replace the .apply function with the vectors:

updated_coordinates = 1/ ( * np.power((np.sqrt(np.power(df_x - gaze_x,2) + np.power(df_y - gaze_y,2))),2))

I've checked and this runs over ten times as fast.

answered Aug 18, 2017 at 10:33

Bram Zijlstra

3901 gold badge5 silver badges13 bronze badges

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

Collectives™ on Stack Overflow

Vectorizing a function using column and row index

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related