Parameters details

For both fastgplearn.skflow.SymbolicRegressor and fastgplearn.skflow.SymbolicClassifier .

Init Parameters:

Parameters name

Type

Default

Suggest Range

Definition

population_size

(int)

10000

[50, 1000000]

number of population

generations

(int)

20

[10,…]

number of generations

tournament_size

(int)

20

[5,20]

number of for each turn of tournament

stopping_criteria

(float)

0.95

[0,1]

early stopping criteria

constant_range

tuple of float

(0,1.0)

/

constants would choice from range randomly

constants

tuple of float

None

/

if given, the parameter constant_range would be ignored, and just use the constants offered

depth

tuple of float

(2,5)

1st [1,…], 2ed [2,8)

(min_depth,max_depth), keep the max of depth is not more than 8 !

function_set

tuple of string

(+-*/)

/

optional: (‘add’, ‘sub’, ‘mul’, ‘div’, “ln”, “exp”, “pow2”, “pow3”, “rec”, “max”, “min”, “sin”, “cos”)

n_jobs

(int)

1

[1,…]

n jobs to parallel

verbose

(bool)

True

True,False

print message

p_mutate

(float)

0.5

(0,1)

mutate probability

p_crossover

(float)

0.5

(0,1)

crossover probability

random_state

(int)

None

/

random state

hall_of_fame

(int)

3

[0,10]

hall of frame number to add to next generation

store_of_fame

(int)

3

[0,10]

hall of frame number to return result

method_backend

(string)

“p_numpy”

/

optional: (“p_numpy”,”c_numpy”,”p_torch”,”c_torch”)

device

(string)

“cpu”

/

optional: (“cpu”,”cuda:0”, …) depend on your computer device

func_p

(np.ndarray)

None

/

specific the probability values of each function

sci_template

list, str

“default”

/

user self-defined list template or “default” or None

Fit Parameters:

Fit parameters in SymbolicRegressor().fit() or SymbolicClassifier().fit() method.

Parameters name

Type

Default

Suggest Range

Definition

X

(np.ndarray)

/

/

with shape (n_sample, n_fea)

y

(np.ndarray)

/

/

with shape (n_sample,)

xs_p

(np.ndarray)

None

/

specific the probability values of each feature

x_label

list of string

None

/

specific the name values of each feature

Other Parameters:

Other parameters present in predict() or score(), or top_n() method.

Parameters name

Type

Default

Suggest Range

Definition

X

(np.ndarray)

/

/

with shape (n_sample, n_fea)

y

(np.ndarray)

/

/

with shape (n_sample,)

n

(int)

0

0

specify the number of expression to calculate or score

scoring

(str)

/

/

for regression, default is “r2”, for classification is “accuracy”