formak.ui_state_machine

Module Contents

Classes

StateId
ConfigView
StateMachineState
NisScore
FitModelState
SymbolicModelState
DesignManager

Data

SearchState
PIPELINE_STAGE_NAME

API

formak.ui_state_machine.SearchState = 'namedtuple(...)'

class formak.ui_state_machine.StateId

Bases: enum.Enum

Start = 0

Symbolic_Model = 'auto(...)'

Fit_Model = 'auto(...)'

class formak.ui_state_machine.ConfigView(params: Dict[str, Any])

Bases: formak.python.Config

Initialization

property common_subexpression_elimination: bool

property python_modules

property extra_validation: bool

property max_dt_sec: float

property innovation_filtering: Optional[float]

class formak.ui_state_machine.StateMachineState(name: str, history: List[formak.ui_state_machine.StateId])

Initialization

abstract classmethod state_id() → formak.ui_state_machine.StateId

history() → List[formak.ui_state_machine.StateId]

abstract classmethod available_transitions() → List[str]

Available function calls.

for each name in the list, getattr(state, name)(*args, **kwargs) will perform the state transition

search(end_state: formak.ui_state_machine.StateId, *, max_iter: int = 100, debug: bool = True) → List[str]

Breadth First Search of state transitions.

For each name in the list, next_state = getattr(current_state, name)(*args, **kwargs) will perform the next state transition

class formak.ui_state_machine.NisScore

__call__(estimator: formak.python.SklearnEKFAdapter, X, y=None) → float

formak.ui_state_machine.PIPELINE_STAGE_NAME = 'kalman'

class formak.ui_state_machine.FitModelState(name: str, history: List[formak.ui_state_machine.StateId], model: formak.ui_model.Model, parameter_space: Dict[str, List[Any]], parameter_sampling_strategy, data, cross_validation_strategy)

Bases: formak.ui_state_machine.StateMachineState

Initialization

classmethod state_id() → formak.ui_state_machine.StateId

classmethod available_transitions() → List[str]

export_python() → formak.python.ExtendedKalmanFilter

_fit_model_impl(debug_print=False)

This impl function contains all of the scikit-learn wrangling to.

organize it away from the logical flow of the state machine. This may move to its own separate helper file.

class formak.ui_state_machine.SymbolicModelState(name: str, history: List[formak.ui_state_machine.StateId], model: formak.ui_model.Model)

Bases: formak.ui_state_machine.StateMachineState

Initialization

classmethod state_id() → formak.ui_state_machine.StateId

classmethod available_transitions() → List[str]

fit_model(parameter_space: Dict[str, List[Any]], data, *, parameter_sampling_strategy=None, cross_validation_strategy=None) → formak.ui_state_machine.FitModelState

Symbolic Model -> Fit Model.

Given the symbolic model contained and the given parameters for hyper-parameter search, perform the hyper-parameter search.

class formak.ui_state_machine.DesignManager(name)

Bases: formak.ui_state_machine.StateMachineState

Initialization

classmethod state_id() → formak.ui_state_machine.StateId

classmethod available_transitions() → List[str]

symbolic_model(model: formak.ui_model.Model) → formak.ui_state_machine.SymbolicModelState