#BP人工神经网络的实现
#1、读取数据
#2、keras.models Sequential /keras.layers.core Dense Activation
#3、Sequential建立模型
#4、Dense建立层
#5、Activation激活函数
#6、compile模型编译
#7、fit训练（学习）
#8、验证（测试，分类预测）
#使用人工神经网络预测课程销量
#数据的读取与整理
import pandas as pda
import numpy as npy
fname = 'D:\shuju\fenleisuanfa\lesson2.csv'
dataf = pda.read_csv(fname)
x = dataf.iloc[:,1:5].values
y = dataf.iloc[:,5:6].values
for i in range(0,len(x)):
for j in range(0,len(x[i])):
thisdata = x[i][j]
if(thisdata =='是' or thisdata == '多' or thisdata == '高'):
x[i][j] = 1
else:
x[i][j] = 0
for i in range(0,len(y)):
thisdata = y[i]
if(thisdata == '高'):
y[i] = 1
else:
y[i] = 0
xf = pda.DataFrame(x)
yf = pda.DataFrame(y)
x2 = xf.values.astype(int)
y2 = yf.values.astype(int)
#使用人工神经网络模型
from keras.models import Sequential
from keras.layers.core import Dense,Activation
import keras.preprocessing.text as t
from keras.preprocessing.text import Tokenizer as tk
from keras.preprocessing.text import text_to_word_sequence
model = Sequential()
#输入层
model.add(Dense(10,input_dim = len(x2[0])))
model.add(Activation('relu'))
#输出层
model.add(Dense(1,input_dim = 1))
model.add(Activation('sigmoid'))
#模型的编译
model.compile(loss = 'binary_crossentropy',optimizer = 'adam', metrics = ['accuracy'])
#训练
rst = model.fit(x2,y2,epochs = 10,batch_size = 100)
#预测分类
model.predict_classes(x).reshape(len(x))
