pflm.fpca.FunctionalPCAUserDefinedParams#

class FunctionalPCAUserDefinedParams(t_mu: ndarray | list[float] = None, mu: ndarray | list[float] = None, t_cov: ndarray | list[float] = None, cov: ndarray | list[list[float]] = None, sigma2: float | None = None, rho: float | None = None)[source][source]#

Bases: object

User-defined parameters for Functional PCA.

Parameters:
t_munp.ndarray or List[float], optional

Time points for the mean function. If provided, must match the length of mu.

munp.ndarray or List[float], optional

Mean function values at the time points in t_mu.

t_covnp.ndarray or List[float], optional

Time points for the covariance function. If provided, must match the dimensions of cov.

covnp.ndarray or List[List[float]], optional

Covariance function values at the time points in t_cov.

sigma2float, optional

Variance of the measurement error.

rhofloat, optional

The user-defined measurement truncation threshold used for conditional expectations estimation on the principal component scores. If provided, must be a non-negative scalar.

Examples

Default (no user-defined overrides):

>>> from pflm.fpca import FunctionalPCAUserDefinedParams
>>> params = FunctionalPCAUserDefinedParams()

Provide a known mean function:

>>> import numpy as np
>>> t_mu = np.linspace(0, 10, 51)
>>> mu = np.sin(t_mu) * 0.5
>>> params = FunctionalPCAUserDefinedParams(t_mu=t_mu, mu=mu)

Fix the measurement error variance:

>>> params = FunctionalPCAUserDefinedParams(sigma2=0.01)