-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathmain.py
201 lines (157 loc) · 7.87 KB
/
main.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
from deap import creator, base, tools
import geppy as gep
import operator
import numpy as np
import sklearn as sk
import math
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import mean_squared_error
from sklearn.metrics import accuracy_score
from sklearn.preprocessing import StandardScaler
from matplotlib import pyplot as plt
from sympy import *
import sympy as sp
from sklearn.metrics import mean_squared_error, r2_score
class Potion:
def __init__(self, ml_model, head=7, genes_q=2, rnc_len=10, pop=120, gener=50, num_best_indv=3):
self.model = ml_model
self.saved_model = None
self.best_ind = None
self.log = None
self.head = head
self.genes_quantity = genes_q
self.rnc_len = rnc_len
self.population = pop
self.generations = gener
self.number_of_best_individs = 3
self.pset = gep.PrimitiveSet('Main', input_names=features.columns.values)
self.pset.add_function(operator.add, 2)
self.pset.add_function(operator.sub, 2)
self.pset.add_function(operator.mul, 2)
self.pset.add_function(self.__protected_div, 2)
self.pset.add_function(self.__protected_mod, 2)
self.pset.add_function(self.__protected_exp, 1)
self.pset.add_function(self.__protected_ln, 1)
self.pset.add_rnc_terminal()
creator.create("FitnessMax", base.Fitness, weights=(1,))
creator.create("Individual", gep.Chromosome, fitness=creator.FitnessMax)
self.toolbox = gep.Toolbox()
self.toolbox.register('rnc_gen', random.randint, a=-10, b=10)
self.toolbox.register('gene_gen', gep.GeneDc, pset=self.pset, head_length=HEAD, rnc_gen=self.toolbox.rnc_gen,
rnc_array_length=self.rnc_len)
self.toolbox.register('individual', creator.Individual, gene_gen=self.toolbox.gene_gen, n_genes=self.genes_quantity,
linker=operator.add)
self.toolbox.register("population", tools.initRepeat, list, self.toolbox.individual)
self.toolbox.register('compile', gep.compile_, pset=self.pset)
self.toolbox.register('select', tools.selTournament, tournsize=3)
self.toolbox.register('evaluate', self.__evaluate)
self.toolbox.register('mut_uniform', gep.mutate_uniform, pset=self.pset, ind_pb=0.05, pb=1)
self.toolbox.register('mut_invert', gep.invert, pb=0.1)
self.toolbox.register('mut_is_transpose', gep.is_transpose, pb=0.1)
self.toolbox.register('mut_ris_transpose', gep.ris_transpose, pb=0.1)
self.toolbox.register('mut_gene_transpose', gep.gene_transpose, pb=0.1)
self.toolbox.register('cx_1p', gep.crossover_one_point, pb=0.3)
self.toolbox.register('cx_2p', gep.crossover_two_point, pb=0.2)
self.toolbox.register('cx_gene', gep.crossover_gene, pb=0.1)
self.toolbox.register('mut_dc', gep.mutate_uniform_dc, ind_pb=0.05, pb=1)
self.toolbox.register('mut_invert_dc', gep.invert_dc, pb=0.1)
self.toolbox.register('mut_transpose_dc', gep.transpose_dc, pb=0.1)
self.toolbox.register('mut_rnc_array_dc', gep.mutate_rnc_array_dc, rnc_gen=self.toolbox.rnc_gen, ind_pb='0.5p')
self.toolbox.pbs['mut_rnc_array_dc'] = 1
self.stats = tools.Statistics(key=lambda ind: ind.fitness.values[0])
self.stats.register("avg", np.mean)
self.stats.register("min", np.min)
self.stats.register("max", np.max)
self.__dataX = None
self.__dataY = None
self.__ml_args = None
self.__metric = None
def __protected_div(self, x1, x2): # we exclude division by zero
try:
x1 / x2
except ZeroDivisionError:
return 1
return x1 / x2
def __protected_mod(self, x1, x2):
try:
x1 % x2
except ZeroDivisionError:
return 1
return x1 % x2
def __protected_exp(self, x):
if x > 10 or abs(x) < 1e6:
return 1
return math.exp(x)
def __protected_ln(self, x):
if x < 0:
x *= -1
if x == 0:
x = 1
return math.log(x)
def fit(self, X_train, Y_train, x_test, y_test, metric=sk.metrics.accuracy_score, **kwargs):
self.__dataX = [X_train, x_test]
self.__dataY = [Y_train, y_test]
self.__ml_args = kwargs
self.__metric = metric
pop = self.toolbox.population(n=self.population)
hof = tools.HallOfFame(self.number_of_best_individs)
pop, log = gep.gep_simple(pop, self.toolbox, n_generations=self.generations, n_elites=1,
stats = self.stats, hall_of_fame = hof, verbose = True)
self.best_ind = hof[0]
func = self.toolbox.compile(self.best_ind) # converting a tree to an expression
x_new_train = np.array(
list(map(func, *[self.__dataX[0][f"{x}"] for x in self.__dataX[0].columns.values]))) # substituting values
x_new_train = x_new_train.reshape(-1, 1)
transformer = StandardScaler().fit(x_new_train)
x_new_train = transformer.transform(x_new_train)
x_new_train = x_new_train.reshape(-1, 1)
ml = self.model # launching the model
ml.fit(np.column_stack([self.__dataX[0], x_new_train]), self.__dataY[0], **self.__ml_args)
self.saved_model = ml
self.log = log
print("Mean squared error: %.2f" % mean_squared_error(y_test, self.predict(x_test)))
print("R2 score : %.2f" % r2_score(y_test, self.predict(x_test)))
return pop, log, hof
def __evaluate(self, individual):
func = self.toolbox.compile(individual) # converting a tree to an expression
x_new_train = np.array(
list(map(func, *[self.__dataX[0][f"{x}"] for x in self.__dataX[0].columns.values]))) # substituting values
x_new_test = np.array(list(map(func, *[self.__dataX[1][f"{x}"] for x in self.__dataX[1].columns.values])))
x_new_train = x_new_train.reshape(-1, 1)
transformer = StandardScaler().fit(x_new_train)
x_new_train = transformer.transform(x_new_train)
x_new_train = x_new_train.reshape(-1, 1)
x_new_test = x_new_test.reshape(-1, 1)
x_new_test = transformer.transform(x_new_test)
ml = self.model # launching the model
ml.fit(np.column_stack([self.__dataX[0], x_new_train]), self.__dataY[0], **self.__ml_args)
pred = ml.predict(np.column_stack([self.__dataX[1], x_new_test]))
accuracy = self.__metric(self.__dataY[1], pred) # evaluating the accuracy
return accuracy,
def predict(self, data):
func = self.toolbox.compile(self.best_ind) # converting a tree to an expression
data_transformed = np.array(
list(map(func, *[data[f"{x}"] for x in data.columns.values]))) # substituting values
data_transformed = data_transformed.reshape(-1, 1)
return self.saved_model.predict(np.column_stack([data, data_transformed]))
def visualize(self):
val_av = []
for i in range(len(self.log)):
val_av.append(self.log[i]['avg'])
val_max = []
for i in range(len(self.log)):
val_max.append(self.log[i]['max'])
plt.figure()
plt.subplot(2, 1, 1)
plt.plot(range(self.generations+1), val_av, label='average', color='green')
plt.ylabel('accuracy')
plt.title("Population Accuracy Change")
plt.legend();
plt.subplot(2, 1, 2)
plt.plot(range(self.generations+1), val_max, label='maximum', color='red')
plt.xlabel('number of generations')
plt.ylabel('accuracy')
plt.legend();
plt.savefig('accuracy_plot.png')
rename_labels = {'add': '+', 'sub': '-', 'mul': '*', '__protected_div': '/', '__protected_ln': 'ln', '__protected_mod': '%', '__protected_exp': 'exp'}
gep.export_expression_tree(self.best_ind, rename_labels, 'numerical_expression_tree.png')