def _normalize_item_avg(input_matrix):
output_matrix = np.copy(input_matrix)
item_sums = np.sum(input_matrix,axis=0)
number_of_users_that_rated_each_item = np.apply_along_axis(qtty_nonzeros,0,input_matrix)
item_avgs = item_sums / number_of_users_that_rated_each_item
for i in range(0,input_matrix.shape[1]):
output_matrix[:,i] = subtract_nonzeros(input_matrix[:,i],item_avgs[i])
return(output_matrix)
def _normalize_user_avg(input_matrix):
output_matrix = np.copy(input_matrix)
user_sums = np.sum(input_matrix,axis=1)
number_of_items_rated_by_each_user = np.apply_along_axis(qtty_nonzeros,1,input_matrix)
user_avgs = user_sums / number_of_items_rated_by_each_user
for i in range(0,input_matrix.shape[0]):
output_matrix[i,:] = subtract_nonzeros(input_matrix[i,:],user_avgs[i])
return(output_matrix)
def _normalize_item_avg(input_matrix):
output_matrix = np.copy(input_matrix)
item_sums = np.sum(input_matrix, axis=0)
number_of_users_that_rated_each_item = np.apply_along_axis(
qtty_nonzeros, 0, input_matrix)
item_avgs = item_sums / number_of_users_that_rated_each_item
for i in range(0, input_matrix.shape[1]):
output_matrix[:, i] = subtract_nonzeros(input_matrix[:, i],
item_avgs[i])
return (output_matrix)
def _normalize_user_avg(input_matrix):
output_matrix = np.copy(input_matrix)
user_sums = np.sum(input_matrix, axis=1)
number_of_items_rated_by_each_user = np.apply_along_axis(
qtty_nonzeros, 1, input_matrix)
user_avgs = user_sums / number_of_items_rated_by_each_user
for i in range(0, input_matrix.shape[0]):
output_matrix[i, :] = subtract_nonzeros(input_matrix[i, :],
user_avgs[i])
return (output_matrix)
