import numpy as np
weight =[0,0,0]
rate = 0.1
def sign(value):
      """符号函数"""
      if value>0:
         return 1
      elif value<0:
         return -1
      else:
         return 0

def crDataSet(datafile):
     """创建数据集""" 
     result = []
     with open(datafile,'r') as file:   
         data = file.read().splitlines()
         for line in data:
              temp = []
              items = line.split(',')
              temp.append(float(items[0]))
              temp.append(float(items[2]))
              if items[4] =='Iris-setosa':
                  temp.append(1)
              else:
                  temp.append(-1)   
              result.append(temp)
     return result
def dotProduct(item):
      """计算点积"""
      sum = weight[0]*1
      for i in range(2):
        sum += item[i]*weight[i+1]
      return sum

def upWeight(item,estimate):
      """更新权重"""
      weight[0] +=  rate*(item[2]-estimate) 
      for i in range(2):
          weight[i+1] +=  rate*(item[2]-estimate)*item[i]
          
            
def train():
    error =1                             #是否错误  
    while error:
          error = 0       
          for item in train_data:
             outvalue = sign(dotProduct(item))   
             while outvalue !=item[2]:
                 error = 1  
                 upWeight(item,outvalue)
                 outvalue = sign(dotProduct(item))  
def test():
    error = 0   
    for item in test_data:              #用测试集集数据进行测试
         if sign(dotProduct(item)) != item[2]:
           error +=1
    print("错误数：",error)
    print(weight)  #查看权重

def disp():
    import matplotlib.pyplot as plt
    plt.xlabel('花萼长度',fontproperties = 'SimHei',fontsize = 12)
    plt.ylabel('花瓣长度',fontproperties = 'SimHei',fontsize = 12)
    positive_x1 = [train_data[i][0] for i in range(60) if train_data[i][2] == 1]
    positive_x2 = [train_data[i][1] for i in range(60) if train_data[i][2] == 1]
    negetive_x1 = [train_data[i][0] for i in range(60) if train_data[i][2] == -1]
    negetive_x2 = [train_data[i][1] for i in range(60) if train_data[i][2] == -1]
    plt.scatter(positive_x1,positive_x2,s=5,c='red')
    plt.scatter(negetive_x1,negetive_x2,s=5,c='blue')
    line_x = np.arange(0,10)
    line_y = (-weight[1]*line_x -weight[0]) / weight[2]
    plt.plot(line_x,line_y)
    plt.show()  

if __name__ =="__main__":
    train_data = crDataSet('train.data')
    test_data = crDataSet('test.data')
    train()
    test()
    disp()
    print(sign(dotProduct([2.2,3.3])))