Documentation for train.py

mylib.train

Module description

SyntheticBernuliDataset(n: int = 10, m: int = 100, seed: int = 42)

Bases: object

Base class for synthetic dataset.

Parameters:

Name	Type	Description	Default
n	int	feature number. Defaults to 10.	10
m	int	object number. Defaults to 100.	100
seed	int	random state seed. Defaults to 42.	42

w = rs.randn(n) instance-attribute

X = rs.randn(m, n) instance-attribute

y = rs.binomial(1, expit(self.X @ self.w)) instance-attribute

Trainer(model, X: np.ndarray, Y: np.ndarray, seed: int = 42)

Bases: object

Base class for all trainers.

Parameters:

Name	Type	Description	Default
model		The class with fit and predict methods.	required
X	ndarray	The array of shape [num_elemennts, num_feature]	required
Y	ndarray	[num_elements, num_answers]	required
seed	int	random state seed. Defaults to 42.	42

model = model instance-attribute

seed = seed instance-attribute

train()

Train model

eval(output_dict: bool = False) -> str | dict

Evaluate model for initial validadtion dataset.

Parameters:

Name	Type	Description	Default
output_dict	bool	If True, return output as dict.	False

Returns:

Type	Description
str | dict	classification report

test(X: np.ndarray, Y: np.ndarray, output_dict: bool = False) -> str | dict

Evaluate model for given dataset.

Parameters:

Name	Type	Description	Default
X	ndarray	The array of shape [num_elements, num_feature]	required
Y	ndarray	The array of shape [num_elements, num_answers]	required
output_dict	bool	If True, return output as dict. Defaults to False.	False

Returns:

Type	Description
str | dict	classification report

cv_parameters(X: np.ndarray, Y: np.ndarray, seed: int = 42, minimal: float = 0.1, maximum: float = 25, count: int = 100) -> tuple[np.ndarray, list[float], list]

Function for the experiment with different regularisation parameters ("Cs") and return accuracy and params for LogisticRegression for each parameter.

Parameters:

Name	Type	Description	Default
X	ndarray	The array of shape [num_elements, num_feature]	required
Y	ndarray	The array of shape [num_elements, num_answers]	required
seed	int	Seed for random state. Defaults to 42.	42
minimal	float	Minimum value for the Cs linspace. Defaults to 0.1.	0.1
maximum	float	Maximum value for the Cs linspace. Defaults to 25.	25
count	int	Number of the Cs points. Defaults to 100.	100

Returns:

Type	Description
ndarray	Cs
list[float]	list of accuracies
list	list of params