遗传算法和粒子群算法简介

mport numpy as np

from sklearn.datasets import make\_classification

from sklearn.svm import SVC

from sklearn.model\_selection import train\_test\_split

from sklearn.metrics import accuracy\_score

from deap import base, creator, tools, algorithms



# 创建适应度评估函数

def evaluate(individual, X, y):

    gamma = individual[0]

    C = individual[1]

    clf = SVC(gamma=gamma, C=C)

    scores = cross\_val\_score(clf, X, y, cv=5)

    return np.mean(scores),



# 创建适应度评估函数的最小化目标

creator.create("FitnessMin", base.Fitness, weights=(-1.0,))



# 创建个体和群体类

creator.create("Individual", list, fitness=creator.FitnessMin)

toolbox = base.Toolbox()

toolbox.register("attr\_float", np.random.uniform, 0, 5)  # 随机生成参数范围在0到5之间的浮点数

toolbox.register("individual", tools.initRepeat, creator.Individual, toolbox.attr\_float, n=2)

toolbox.register("population", tools.initRepeat, list, toolbox.individual)

toolbox.register("mate", tools.cxBlend, alpha=0.5)  # 交叉算子

toolbox.register("mutate", tools.mutGaussian, mu=0, sigma=0.2, indpb=0.2)  # 变异算子

toolbox.register("select", tools.selTournament, tournsize=3)  # 选择算子

toolbox.register("evaluate", evaluate)



# 生成示例数据

X, y = make\_classification(n\_samples=1000, n\_features=20, n\_classes=2, random\_state=42)



# 划分训练集和测试集

X\_train, X\_test, y\_train, y\_test = train\_test\_split(X, y, test\_size=0.2, random\_state=42)



# 生成初始种群

population = toolbox.population(n=50)



# 迭代进化

NGEN = 10

for gen in range(NGEN):

    offspring = algorithms.varAnd(population, toolbox, cxpb=0.5, mutpb=0.1)

    fits = toolbox.map(toolbox.evaluate, offspring)

    for fit, ind in zip(fits, offspring):

        ind.fitness.values = fit

    population = toolbox.select(offspring, k=len(population))



# 输出最优解

best\_individual = tools.selBest(population, k=1)[0]

best\_gamma, best\_C = best\_individual

print("Best gamma:", best\_gamma)

print("Best C:", best\_C)



# 使用最优参数训练模型

clf = SVC(gamma=best\_gamma, C=best\_C)

clf.fit(X\_train, y\_train)



# 在测试集上评估模型性能

y\_pred = clf.predict(X\_test)

accuracy = accuracy\_score(y\_test, y\_pred)

print("Test Accuracy:", accuracy)

import numpy as np

from sklearn.datasets import make\_classification

from sklearn.svm import SVC

from sklearn.model\_selection import train\_test\_split

from sklearn.metrics import accuracy\_score

from pyswarm import pso



# 生成示例数据

X, y = make\_classification(n\_samples=1000, n\_features=20, n\_classes=2, random\_state=42)



# 划分训练集和测试集

X\_train, X\_test, y\_train, y\_test = train\_test\_split(X, y, test\_size=0.2, random\_state=42)



# 定义适应度评估函数

def evaluate(parameters, X, y):

    gamma, C = parameters

    clf = SVC(gamma=gamma, C=C)

    clf.fit(X, y)

    y\_pred = clf.predict(X)

    accuracy = accuracy\_score(y, y\_pred)

    return -accuracy  # 负号表示最小化问题



# 定义参数优化的目标函数

def objective\_function(parameters):

    return evaluate(parameters, X\_train, y\_train)



# 设定参数范围

lb = [0.1, 0.1]  # 参数下界

ub = [10, 10]    # 参数上界



# 使用粒子群算法进行优化

parameters, \_ = pso(objective\_function, lb, ub, swarmsize=10, maxiter=20)



# 输出优化后的最优参数

best\_gamma, best\_C = parameters

print("Best gamma:", best\_gamma)

print("Best C:", best\_C)



# 使用最优参数训练模型

clf = SVC(gamma=best\_gamma, C=best\_C)

clf.fit(X\_train, y\_train)



# 在测试集上评估模型性能

y\_pred = clf.predict(X\_test)

accuracy = accuracy\_score(y\_test, y\_pred)

print("Test Accuracy:", accuracy)