一、数据来源

1. 数据来源：UCI

1. 数据样式

最后一列数据为判断是否为物体，其他列为参数数值。

二、使用方法

感知机（perception）

感知机是一个线性分类器，它是最简单的神经网络，只有一个神经元，通过结果的正负来做二分类。

通过梯度下降，对系数不断优化，寻找最优系数:

三、代码实现

从数据读取开始，不调取三方库，纯手工推。

1. 导入基础库

from random import seed,randrange
from csv import reader

1. 读取csv文件和转换字符串为浮点型

def csv_loader(file):dataset=list()with open(file,'r') as f:csv_reader =reader(f)for row in csv_reader:if not row:continuedataset.append(row)return datasetdef str_to_float_converter(dataset):for i in range(len(dataset[0])-1):for row in dataset:row[i]=float(row[i].strip())

1. 原数据集最后一列转为0或1

def last_column_to_int(dataset):col_value=[row[-1] for row in dataset]uni_value=set(col_value)search_tool=dict()for i,value in enumerate(uni_value):search_tool[value]=ifor row in dataset:row[-1]= search_tool[row[-1]]

1. 使用k-fold cross validation拆分数据

def k_folds_cross_validation(dataset,n_folds):dataset_split =list()dataset_copy=list(dataset)fold_size=int(len(dataset)/n_folds)for i in range(n_folds):fold = list()while len(fold) < fold_size:index = randrange(len(dataset_copy))fold.append(dataset_copy.pop(index))dataset_split.append(fold)return dataset_split

1. 计算准确性

def calculate_accuracy(actual,predicted):correct =0for i in range(len(actual)):if actual[i] == predicted[i]:correct+=1return correct/float(len(actual)) * 100.0

1. 模型测试

使用K折交叉验证，评估每一次折叠的模型准确性。

其中：algo为算法占位，具体算法写好后，在运行时迭代即可。

def model_test(dataset,algo,n_folds,*args):folds=k_folds_cross_validation(dataset,n_folds)scores =list()for fold in folds:train=list(folds)train.remove(fold)train=sum(train,[])test=list()for row in fold:row_copy=list(row)test.append(row_copy)row_copy[-1]=Nonepredicted = algo(train,test,*args)actual =[row[-1] for row in fold]accuracy = calculate_accuracy(actual,predicted)scores.append(accuracy)return scores

1. 预测数据的基础模型

感知机本质上是一个线性模型，通过结果的正负来做二分类

def predict(row,weights):activation = weights[0]for i in range(len(row)-1):activation += weights[i+1] *row[i]return 1.0 if activation >= 0.0 else 0

1. 使用随机梯度下降法预估系数

def estimats_weights(train,learning_rate,n_epochs):weights=[0.0 for i in range(len(train[0]))]for epoch in range(n_epochs):sse=0for row in train:prediction = predict(row,weights)error = prediction - row[-1]sse = error **2weights[0] = weights[0] - learning_rate * errorfor i in range(len(row)-1):weights[i+1] = weights[i+1] - learning_rate * error * row[i]print('This is epoch < %s >, sum_error is < %.4f >' %(epoch,sse))return weights

1. 感知机函数，用来预测测试数据

def perception(train,test,learning_rate,n_epochs):predictions = list()weights = estimats_weights(train,learning_rate,n_epochs)for row in test:prediction = predict(row,weights)predictions.append(prediction)return predictions

1. 运行和参数调整

seed(1)
file='./download_datas/sonar.all-data.csv'
dataset=csv_loader(file)
str_to_float_converter(dataset)
last_column_to_int(dataset)n_folds=5
learning_rate=0.0001
n_epochs=12
algo=perceptionscores=model_test(dataset,algo,n_folds,learning_rate,n_epochs)
print('The score of my model are : %s ' %scores)
print('The average accuracy is : %.3f%% ' %(sum(scores)/float(len(scores))))

运行结果：

The score of my model are : [80.48780487804879, 82.92682926829268, 73.17073170731707, 82.92682926829268, 70.73170731707317]
The average accuracy is : 78.049% 

四、完整代码

#1.导入基础库
from random import seed,randrange
from csv import reader#2.读取csv文件和转换字符串为浮点型
def csv_loader(file):dataset=list()with open(file,'r') as f:csv_reader =reader(f)for row in csv_reader:if not row:continuedataset.append(row)return datasetdef str_to_float_converter(dataset):for i in range(len(dataset[0])-1):for row in dataset:row[i]=float(row[i].strip())#3.原数据集最后一列转为0或1
def last_column_to_int(dataset):col_value=[row[-1] for row in dataset]uni_value=set(col_value)search_tool=dict()for i,value in enumerate(uni_value):search_tool[value]=ifor row in dataset:row[-1]= search_tool[row[-1]]#4.使用k_folds cross validation(交叉确认)
def k_folds_cross_validation(dataset,n_folds):dataset_split =list()dataset_copy=list(dataset)fold_size=int(len(dataset)/n_folds)for i in range(n_folds):fold = list()while len(fold) < fold_size:index = randrange(len(dataset_copy))fold.append(dataset_copy.pop(index))dataset_split.append(fold)return dataset_split#5.计算准确性
def calculate_accuracy(actual,predicted):correct =0for i in range(len(actual)):if actual[i] == predicted[i]:correct+=1return correct/float(len(actual)) * 100.0#6.whether the algo is good or not?
def model_test(dataset,algo,n_folds,*args):folds=k_folds_cross_validation(dataset,n_folds)scores =list()for fold in folds:train=list(folds)train.remove(fold)train=sum(train,[])test=list()for row in fold:row_copy=list(row)test.append(row_copy)row_copy[-1]=Nonepredicted = algo(train,test,*args)actual =[row[-1] for row in fold]accuracy = calculate_accuracy(actual,predicted)scores.append(accuracy)return scores#7.预测的模型
def predict(row,weights):activation = weights[0]for i in range(len(row)-1):activation += weights[i+1] *row[i]return 1.0 if activation >= 0.0 else 0#8.使用随机梯度下降法预估系数
def estimats_weights(train,learning_rate,n_epochs):weights=[0.0 for i in range(len(train[0]))]for epoch in range(n_epochs):sse=0for row in train:prediction = predict(row,weights)error = prediction - row[-1]sse = error **2weights[0] = weights[0] - learning_rate * errorfor i in range(len(row)-1):weights[i+1] = weights[i+1] - learning_rate * error * row[i]print('This is epoch < %s >, sum_error is < %.4f >' %(epoch,sse))return weights#9.感知机函数，用来预测测试数据
def perception(train,test,learning_rate,n_epochs):predictions = list()weights = estimats_weights(train,learning_rate,n_epochs)for row in test:prediction = predict(row,weights)predictions.append(prediction)return predictions#10.运行和参数调整
seed(1)
file='./download_datas/sonar.all-data.csv'
dataset=csv_loader(file)
str_to_float_converter(dataset)
last_column_to_int(dataset)n_folds=5
learning_rate=0.0001
n_epochs=12
algo=perceptionscores=model_test(dataset,algo,n_folds,learning_rate,n_epochs)
print('The score of my model are : %s ' %scores)
print('The average accuracy is : %.3f%% ' %(sum(scores)/float(len(scores))))