feat: add genetic algorithm library

Python_codes/Genetic Algorithm/ga.py

@@ -0,0 +1,96 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+#
+# Copyright 2021 Jude Michael Teves
+#
+# This file is a modification of the genetic algorithm library made by the same author.
+# The link of the original implementation can be found here: https://github.com/Cyntwikip/libraries/blob/master/genetic_algo/ga.py
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, see <http://www.gnu.org/licenses/>.
+
+import ast, random, itertools
+
+import numpy as np 
+import pandas as pd 
+
+from collections.abc import Iterable
+
+class GeneticAlgo():
+
+    def __init__(self, genes_dict, debug=False, random_seed=42):
+        self.debug = debug
+        self.random_seed = random_seed
+        self.genes_dict = genes_dict
+        self.generate_chromosomes()
+
+    def generate_chromosomes(self):
+        keys = self.genes_dict.keys()
+        values = self.genes_dict.values()
+        chromosomes = list(itertools.product(*values))
+        c_len = len(chromosomes)
+        codes = [str(i).zfill(5) for i in range(c_len)]
+        chromosomes = pd.DataFrame(chromosomes, index=codes, columns=keys)
+        self.chromosomes = chromosomes
+        if self.debug:
+            print('Number of Chromosomes: {}'.format(chromosomes.shape[0]))
+
+    def get_initial_population(self, n):
+        np.random.seed(self.random_seed)
+        codes = list(self.chromosomes.index)
+        return np.random.choice(codes, n, replace=False)
+
+    def get_best_parents(self, fitness):
+        '''
+        fitness - dict {code: score}
+        '''
+        result = sorted(fitness.items(), key=lambda x: x[1], reverse=True)[:2]
+        result = [i[0] for i in result] # get code
+        # duplicate single parent
+        if len(result) == 1:
+            result = result*2
+        return result
+
+    def make_children(self, parent1_code, parent2_code, num_children, mutation_prob=0.1):
+        # random seed
+        np.random.seed(self.random_seed)
+        random.seed(self.random_seed)
+
+        parent_prob = (1 - mutation_prob)/2
+        children_chromosomes = []
+        for i in range(num_children):
+            chromosome = []
+            for key in self.chromosomes:
+                choices = [self.chromosomes.loc[parent1_code, key], 
+                            self.chromosomes.loc[parent2_code, key],
+                            random.choice(self.genes_dict[key])]              
+                if type(choices[0]) != str and isinstance(choices[0], Iterable):
+                    choices = [str(i) for i in choices]
+                    gene = [np.random.choice(choices, p = [parent_prob, parent_prob, mutation_prob])][0]
+                    gene = ast.literal_eval(gene)
+                else:
+                    gene = [np.random.choice(choices, p = [parent_prob, parent_prob, mutation_prob])][0]
+
+                chromosome.append(gene)
+            children_chromosomes.append(chromosome)
+
+        codes = list(map(self.get_chromosome_code, children_chromosomes))
+        codes = list(set(codes))
+        return codes
+
+    def get_chromosome_code(self, chromosome):
+        c = self.chromosomes
+        for key, row in c.iterrows():
+            if (row.values.tolist()==chromosome):        
+                break
+        return key

Python_codes/Genetic Algorithm/sample.py

@@ -0,0 +1,112 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+#
+# Copyright 2021 Jude Michael Teves
+#
+# This file is a modification of the genetic algorithm library made by the same author.
+# The link of the original implementation can be found here: https://github.com/Cyntwikip/libraries/blob/master/genetic_algo/sample.py
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, see <http://www.gnu.org/licenses/>.
+
+import click
+from ga import GeneticAlgo as ga
+
+import numpy as np
+
+# Pool of genes
+pool_scaler = ["MinMax (-1,1)", "MinMax (0,1)", "Std"]
+pool_batch_size = [64, 72, 128, 256]
+pool_lstm_layers = [[128, 128, 128],
+                    [64, 64, 64],
+                    [32, 32, 32],
+                    [128, 64, 64],
+                    [64, 32, 32],
+                    [128, 64, 32]]
+pool_dropout = [0., 0.1, 0.2]
+pool_activation = ['tanh', 'sigmoid']
+pool_loss = ['mean_squared_error', 'mae']
+pool_epochs = [5]
+
+genes_dict = dict( scaler = pool_scaler,
+                batch_size = pool_batch_size,
+                lstm_layers = pool_lstm_layers,
+                dropout = pool_dropout,
+                activation = pool_activation,
+                loss = pool_loss,
+                epochs = pool_epochs
+            )
+
+ga = ga(genes_dict, debug=True)
+
+@click.group()
+def cli():
+    pass
+
+@cli.command(name='make-children')
+@click.argument('code1', type=str)
+@click.argument('code2', type=str)
+@click.option('-n', default=10)
+def make_children(code1, code2, n):
+    codes = ga.make_children(code1, code2, n)
+    print(codes)
+    return
+
+@cli.command(name='chromosomes-head')
+@click.option('-n', default=5)
+def chromosomes_head(n):
+    print(ga.chromosomes.head(n))
+
+@cli.command(name='get-population')
+@click.option('-n', default=10)
+def get_population(n):
+    pop = ga.get_initial_population(n)
+    print(pop)
+
+@cli.command(name='get-chromosome')
+@click.argument('code', type=str)
+def get_chromosome(code):
+    print(ga.chromosomes.loc[code])
+
+@cli.command(name='survival')
+@click.option('-n', default=10)
+def survival(n):
+    # n=1
+    codes = ga.get_initial_population(n)
+    np.random.seed(42)
+    scores = np.random.randint(0,40,n)
+    fitness = dict(zip(codes, scores))
+    print(fitness)
+    best_parents = ga.get_best_parents(fitness)
+    print(best_parents)
+
+@cli.command(name='random')
+@click.option('-n', default=5)
+def randomizer(n):
+    np.random.seed(42)
+    a = [1,2,3,4,5,6,7,8,9,10]
+
+    for i in range(n):
+        print(np.random.choice(a))
+
+@cli.command(name='test1')
+def test1():
+    import pandas as pd
+    df = pd.DataFrame({'a':[[0,1,3],[2,3,4]], 'b':['a','b'], 'c':[64, 128]})
+    compare = [[0,1,3], 'a', 64]
+    for key, row in df.iterrows():    
+        if (row.values==compare).all():
+            print(key)
+
+if __name__ == '__main__':
+    cli()