shamboflow.engine.losses

module documentation
(source)

A small collection of loss functions

Function get Helper function to get a loss function
Function MSE Mean Squared Error
def get(func: str): (source)

Helper function to get a loss function

Return the appropriate loss function depending on the given string

Args
func:
Query string for the requested loss function
Returns
Appropriate function
def MSE(pred_val: np.ndarray, obs_val: np.ndarray) -> float: (source)

Mean Squared Error

It is a loss/cost function to calculate error between the predicted data and the actual data.

MSE = (1 / 2)(Sum (i = 1 to n)(y_i - y^_i) ^ 2)

It is the mean of all errors squared. It is multiplied with 1/2 instead of 1/n to simplify the partial derivative.

Args
pred_val:
The input vector to calculate the error for
obs_val:
The actual observed values
Returns
Overall error of the output