Models¶
- rascal.models.poly_cost_function(a, x, y, degree)[source]¶
Polynomial cost function. Returns the absolute difference between the target value and predicted values.
- Parameters
a (list) – Polynomial coefficients
x (list) – Values to evaluate polynomial at
y (list) – Target values for each x
degree (int) – Polynomial degree
- Returns
residual – y - f(x)
- Return type
list
- rascal.models.polynomial(a, degree=3)[source]¶
Returns a lambda function which computes an nth order polynormal:
f(x, a) = sum_i (a[degree-i] * x**i )
- rascal.models.robust_polyfit(x, y, degree=3, x0=None)[source]¶
Perform a robust polyfit given a set of values (x,y).
Specifically this function performs a least squares fit to the given data points using the robust Huber loss. Inputs are normalised prior to fitting.
- Parameters
x (list) – Data points
y (list) – Target data to fit
degree (int) – Polynomial degree to fit
x0 (list or None) – Initial coefficients
- Returns
p – Polynomial coefficients
- Return type
list