The inner ML algorithm is now sent as an argument

Author: jespb

Arguments.py

@@ -1,11 +1,15 @@
 from sys import argv
 
+from m3gp.MahalanobisDistanceClassifier import MahalanobisDistanceClassifier
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.tree import DecisionTreeRegressor
+
 # 
 # By using this file, you are agreeing to this product's EULA
 #
 # This product can be obtained in https://github.com/jespb/Python-M3GP
 #
-# Copyright ©2019-2022 J. E. Batista
+# Copyright ©2019-2025 J. E. Batista
 #
 
 
@@ -58,7 +62,8 @@
 RANDOM_STATE = 42
 
 # Models wrapped by the M3GP models
-MODEL_NAME = ["MahalanobisDistanceClassifier", "RandomForestClassifier", "DecisionTreeRegressor"][0]
+MODEL = [MahalanobisDistanceClassifier(), RandomForestClassifier(max_depth=6), DecisionTreeRegressor(max_depth=6)][0]
+MODEL_NAME = MODEL.__class__.__name__
 
 # Fitness used by the M3GP models
 FITNESS_TYPE = ["Accuracy", "MSE", "WAF", "2FOLD"][0]

Main_M3GP_classification_example.py

@@ -1,6 +1,7 @@
 import pandas
 
 from m3gp.M3GP import M3GP
+from sklearn.ensemble import RandomForestClassifier
 
 from sklearn.model_selection import train_test_split
 
@@ -17,7 +18,7 @@
 #
 # This product can be obtained in https://github.com/jespb/Python-M3GP
 #
-# Copyright ©2019-2021 J. E. Batista
+# Copyright ©2019-2025 J. E. Batista
 #
 
 
@@ -33,7 +34,7 @@
 		train_size=0.7, random_state = 42, stratify = ds[class_header])
 
 # Train a model
-m3gp = M3GP()
+m3gp = M3GP(model_class=RandomForestClassifier(max_depth=6), fitnessType="2FOLD")
 m3gp.fit(Tr_X, Tr_Y)
 
 # Predict test results

Main_M3GP_regression_example.py

@@ -6,6 +6,9 @@
 
 from sklearn.metrics import mean_squared_error
 
+
+from sklearn.tree import DecisionTreeRegressor
+
 import warnings
 
 warnings.filterwarnings("ignore", category=FutureWarning,
@@ -17,12 +20,12 @@
 #
 # This product can be obtained in https://github.com/jespb/Python-M3GP
 #
-# Copyright ©2019-2021 J. E. Batista
+# Copyright ©2019-2025 J. E. Batista
 #
 
 
 
-filename= "boom_bikes.csv"
+filename= "heart.csv"
 
 # Open the dataset
 ds = pandas.read_csv("datasets/"+filename)
@@ -33,7 +36,7 @@
 		train_size=0.7, random_state = 42)
 
 # Train a model
-m3gp = M3GP(max_generation=20, model_name="DecisionTreeRegressor", fitnessType="MSE",random_state=21)
+m3gp = M3GP(max_generation=20, model_class=DecisionTreeRegressor(max_depth=5), fitnessType="MSE",random_state=21)
 m3gp.fit(Tr_X, Tr_Y, Te_X, Te_Y)
 
 # Predict test results

Main_M3GP_standalone.py

@@ -21,7 +21,7 @@
 #
 # This product can be obtained in https://github.com/jespb/Python-M3GP
 #
-# Copyright ©2019-2024 J. E. Batista
+# Copyright ©2019-2025 J. E. Batista
 #
 
 
@@ -53,7 +53,7 @@ def run(r,dataset):
 
 	# Train a model
 	m3gp = M3GP(OPERATORS, MAX_DEPTH, POPULATION_SIZE, MAX_GENERATION, TOURNAMENT_SIZE, 
-		ELITISM_SIZE, LIMIT_DEPTH, DIM_MIN, DIM_MAX, THREADS, r, VERBOSE, MODEL_NAME, FITNESS_TYPE)
+		ELITISM_SIZE, LIMIT_DEPTH, DIM_MIN, DIM_MAX, THREADS, r, VERBOSE, MODEL, FITNESS_TYPE)
 	m3gp.fit(Tr_X, Tr_Y, Te_X, Te_Y)
 
 

m3gp/GeneticOperators.py

@@ -6,7 +6,7 @@
 #
 # This product can be obtained in https://github.com/jespb/Python-M3GP
 #
-# Copyright ©2019-2022 J. E. Batista
+# Copyright ©2019-2025 J. E. Batista
 #
 
 
@@ -97,7 +97,7 @@ def STXO(rng, population, tournament_size):
 
 	ret = []
 	for d in [d1,d2]:
-		i = Individual(ind1.operators, ind1.terminals, ind1.max_depth, ind1.model_name, ind1.fitnessType)
+		i = Individual(ind1.operators, ind1.terminals, ind1.max_depth, ind1.model_class, ind1.fitnessType)
 		i.copy(d)
 		ret.append(i)
 	return ret
@@ -126,7 +126,7 @@ def M3XO(rng, population, tournament_size):
 
 	ret = []
 	for d in [d1,d2]:
-		i = Individual(ind1.operators, ind1.terminals, ind1.max_depth, ind1.model_name, ind1.fitnessType)
+		i = Individual(ind1.operators, ind1.terminals, ind1.max_depth, ind1.model_class, ind1.fitnessType)
 		i.copy(d)
 		ret.append(i)
 	return ret
@@ -149,7 +149,7 @@ def STMUT(rng, population, tournament_size):
 
 
 	ret = []
-	i = Individual(ind1.operators, ind1.terminals, ind1.max_depth, ind1.model_name, ind1.fitnessType)
+	i = Individual(ind1.operators, ind1.terminals, ind1.max_depth, ind1.model_class, ind1.fitnessType)
 	i.copy(d1)
 	ret.append(i)
 	return ret
@@ -171,7 +171,7 @@ def M3ADD(rng, population, tournament_size, dim_max):
 		n.create(rng, ind1.operators, ind1.terminals, ind1.max_depth)
 		d1.append(n)
 
-		i = Individual(ind1.operators, ind1.terminals, ind1.max_depth, ind1.model_name, ind1.fitnessType)
+		i = Individual(ind1.operators, ind1.terminals, ind1.max_depth, ind1.model_class, ind1.fitnessType)
 		i.copy(d1)
 		ret.append(i)
 
@@ -193,7 +193,7 @@ def M3REM(rng, population, tournament_size, dim_min):
 		r1 = rng.randint(0,len(d1)-1)
 		d1.pop(r1)
 
-		i = Individual(ind1.operators, ind1.terminals, ind1.max_depth, ind1.model_name, ind1.fitnessType)
+		i = Individual(ind1.operators, ind1.terminals, ind1.max_depth, ind1.model_class, ind1.fitnessType)
 		i.copy(d1)
 		ret.append(i)
 

m3gp/Individual.py

@@ -1,10 +1,9 @@
-from sklearn.tree import DecisionTreeRegressor
 from .Node import Node
-from .MahalanobisDistanceClassifier import MahalanobisDistanceClassifier
-from sklearn.ensemble import RandomForestClassifier
 
 import pandas as pd
 
+from copy import deepcopy
+
 from sklearn.metrics import accuracy_score, f1_score, cohen_kappa_score, mean_squared_error
 
 
@@ -13,7 +12,7 @@
 #
 # This product can be obtained in https://github.com/jespb/Python-M3GP
 #
-# Copyright ©2019-2022 J. E. Batista
+# Copyright ©2019-2025 J. E. Batista
 #
 
 class Individual:
@@ -34,11 +33,11 @@ class Individual:
 
 	model = None
 
-	def __init__(self, operators, terminals, max_depth, model_name="MahalanobisDistanceClassifier", fitnessType="Accuracy"):
+	def __init__(self, operators, terminals, max_depth, model_class=None, fitnessType="Accuracy"):
 		self.operators = operators
 		self.terminals = terminals
 		self.max_depth = max_depth
-		self.model_name = model_name
+		self.model_class = model_class
 		self.fitnessType = fitnessType
 
 	def create(self,rng, n_dims=1):
@@ -74,12 +73,7 @@ def __str__(self):
 
 
 	def createModel(self):
-		if self.model_name == "MahalanobisDistanceClassifier":
-			return MahalanobisDistanceClassifier()
-		elif self.model_name == "RandomForestClassifier":
-			return RandomForestClassifier(random_state = 42, max_depth=6)
-		elif self.model_name == "DecisionTreeRegressor":
-			return DecisionTreeRegressor(random_state = 42, max_depth=6)
+		return deepcopy(self.model_class)
 
 	def fit(self, Tr_x, Tr_y):
 		'''
@@ -150,7 +144,7 @@ def getFitness(self, tr_x = None, tr_y = None):
 			if self.fitnessType == "MSE":
 				self.fit(self.training_X, self.training_Y)
 				self.getTrainingPredictions()
-				mse = -1 * mean_squared_error(self.trainingPredictions, self.training_Y)
+				mse = float(-1 * mean_squared_error(self.trainingPredictions, self.training_Y))
 				self.fitness = mse 
 
 			if self.fitnessType == "WAF":
@@ -319,15 +313,15 @@ def prun(self,min_dim=1,simp=False):
 
 		dup = self.dimensions[:]
 		i = 0
-		ind = Individual(self.operators, self.terminals, self.max_depth, self.model_name, self.fitnessType)
+		ind = Individual(self.operators, self.terminals, self.max_depth, self.model_class, self.fitnessType)
 		ind.copy(dup)
 
 		ind.fit(self.training_X, self.training_Y)
 
 		while i < len(dup) and len(dup) > min_dim:
 			dup2 = dup[:]
 			dup2.pop(i)
-			ind2 = Individual(self.operators, self.terminals, self.max_depth, self.model_name, self.fitnessType)
+			ind2 = Individual(self.operators, self.terminals, self.max_depth, self.model_class, self.fitnessType)
 			ind2.copy(dup2)
 			ind2.fit(self.training_X, self.training_Y)
 

m3gp/M3GP.py

@@ -3,14 +3,16 @@
 import multiprocessing as mp
 import time
 
+from .MahalanobisDistanceClassifier import MahalanobisDistanceClassifier
+
 from random import Random
 
 # 
 # By using this file, you are agreeing to this product's EULA
 #
 # This product can be obtained in https://github.com/jespb/Python-M3GP
 #
-# Copyright ©2019-2022 J. E. Batista
+# Copyright ©2019-2025 J. E. Batista
 #
 
 class ClassifierNotTrainedError(Exception):
@@ -38,7 +40,7 @@ class M3GP:
 	dim_min = None
 	dim_max = None
 
-	model_name = None 
+	model_class = None 
 	fitnessType = None
 
 	verbose = None
@@ -73,7 +75,7 @@ def checkIfTrained(self):
 
 	def __init__(self, operators=[("+",2),("-",2),("*",2),("/",2)], max_initial_depth = 6, population_size = 500, 
 		max_generation = 100, tournament_size = 5, elitism_size = 1, max_depth = 17, 
-		dim_min = 1, dim_max = 9999, threads=1, random_state = 42, verbose = True, model_name="MahalanobisDistanceClassifier", fitnessType="Accuracy"):
+		dim_min = 1, dim_max = 9999, threads=1, random_state = 42, verbose = True, model_class=None, fitnessType="Accuracy"):
 
 		if sum( [0 if op in [("+",2),("-",2),("*",2),("/",2)] else 0 for op in operators ] ) > 0:
 			print( "[Warning] Some of the following operators may not be supported:", operators)
@@ -93,7 +95,9 @@ def __init__(self, operators=[("+",2),("-",2),("*",2),("/",2)], max_initial_dept
 		self.dim_min = max(1, dim_min)
 		self.dim_max = max(1, dim_max)
 
-		self.model_name = model_name
+		self.model_class = model_class
+		if self.model_class is None:
+			self.model_class = MahalanobisDistanceClassifier()
 		self.fitnessType = fitnessType
 
 		self.verbose = verbose
@@ -191,7 +195,7 @@ def fit(self,Tr_x, Tr_y, Te_x = None, Te_y = None):
 			print("    > Max Depth:          "+str(self.max_depth))
 			print("    > Minimum Dimensions: "+str(self.dim_min))
 			print("    > Maximum Dimensions: "+str(self.dim_max))
-			print("    > Wrapped Model:      "+self.model_name)
+			print("    > Wrapped Model:      "+self.model_class.__class__.__name__)
 			print("    > Fitness Type:       "+self.fitnessType)
 			print("    > Threads:            "+str(self.threads))
 			print()
@@ -206,12 +210,13 @@ def fit(self,Tr_x, Tr_y, Te_x = None, Te_y = None):
 		self.population = []
 
 		while len(self.population) < self.population_size:
-			ind = Individual(self.operators, self.terminals, self.max_depth, self.model_name, self.fitnessType)
+			ind = Individual(self.operators, self.terminals, self.max_depth, self.model_class, self.fitnessType)
 			ind.create(self.rng, n_dims = self.dim_min)
 			self.population.append(ind)
 
 		self.bestIndividual = self.population[0]
 		self.bestIndividual.fit(self.Tr_x, self.Tr_y)
+		self.bestIndividual.getFitness()
 
 		if not self.Te_x is None:
 			self.trainingAccuracyOverTime = []

m3gp/MahalanobisDistanceClassifier.py

@@ -5,7 +5,7 @@
 #
 # This product can be obtained in https://github.com/jespb/Python-M3GP
 #
-# Copyright ©2019-2022 J. E. Batista
+# Copyright ©2019-2025 J. E. Batista
 #
 
 #
@@ -77,8 +77,6 @@ def fit(self,X,Y):
 		'''
 		Calculates the class clusters in the output space.
 		'''
-		if self.invCovarianceMatrix != None:
-			return
 
 		X = [ list(sample) for sample in X.iloc ]
 		Y = list(Y)