项目实训 --主观题阅卷评分网站（十二）

项目进度：

根据stacking集成学习算法的思想，划分数据集进行训练。

# coding: utf-8
import sys

sys.path.append('..')  # 为了引入父目录的文件而进行的设定
from common.optimizer import SGD, Adam
from common.trainer import Trainer
from common.base_model import BaseModel
# from dataset import spiral
from two_layer_net import TwoLayerNet
import numpy as np
from readCsv import get_attention_vector_and_label
from yigexiangfa import softmax, softmax1, softmax2, getDim
import torch

# 设定超参数
max_epoch = 1
batch_size = 64
hidden_size = 1024  # 256 | 512 | 768 | 1024
learning_rate = 0.005
TRAINNUM = 10000
TESTNUM = 2000
nest_degree = 4  # 2 | 3 | 4 | 6
outputSize = 3

file_name_input_data = 'input1-1-mode.csv'
file_name_label = 'vec.csv'
FILE_NAME = 'two_layer_net.pkl'

referencce = '无权。根据公司法律制度的规定，名义股东处分股权造成实际出资人损失，实际出资人请求名义股东承担赔偿责任的，人民法院应予以支持'

inputDim = int(2 * getDim(referencce, nest_degree))

x0, t0 = get_attention_vector_and_label(TRAINNUM + TESTNUM, outputSize,
                                        file_name_input_data, file_name_label, nest_degree)
x1 = x0[0:2000]
t1 = t0[0:2000]

x2 = x0[2000:4000]
t2 = t0[2000:4000]

x3 = x0[4000:6000]
t3 = t0[4000:6000]

x4 = x0[6000:8000]
t4 = t0[6000:8000]

x5 = x0[8000:10000]
t5 = t0[8000:10000]

e1 = np.vstack((x2, x3, x4, x5))  # 1
e2 = np.vstack((x1, x3, x4, x5))  # 2
e3 = np.vstack((x1, x2, x4, x5))  # 3
e4 = np.vstack((x1, x2, x3, x5))  # 4
e5 = np.vstack((x1, x2, x3, x4))  # 5

f1 = np.vstack((t2, t3, t4, t5))  # 1
f2 = np.vstack((t1, t3, t4, t5))  # 2
f3 = np.vstack((t1, t2, t4, t5))  # 3
f4 = np.vstack((t1, t2, t3, t5))  # 4
f5 = np.vstack((t1, t2, t3, t4))  # 5
mat = []
ceshi = np.zeros((2000, 3))
for i in range(5):
    model = TwoLayerNet(input_size=inputDim, hidden_size=hidden_size,
                        output_size=outputSize, file_name=FILE_NAME)
    optimizer = SGD(lr=learning_rate)
    # optimizer = Adam(lr=learning_rate)
    if (i == 0):
        x = e1
        t = f1
        yuce = x1
    elif (i == 1):
        x = e2
        t = f2
        yuce = x2
    elif (i == 2):
        x = e3
        t = f3
        yuce = x3
    elif (i == 3):
        x = e4
        t = f4
        yuce = x4
    else:
        x = e5
        t = f5
        yuce = x5
    trainer = Trainer(model, optimizer)
    trainer.fit(x, t, max_epoch, batch_size, eval_interval=10)

    # 保存必要数据，以便后续使用
    # BaseModel.save_params(model, 'two_layer_net.pkl')

    print('||  TEST  || -------------------------------------')

    text_data = x0[10000:12000]
    text_result = model.predict(text_data).tolist()
    ceshi = ceshi + text_result
    text_label = t0[10000:12000].tolist()

    text_data2 = yuce
    text_result2 = model.predict(text_data2)
    ACR2=0
    for text_i in range(0, len(text_result2)):
        maxindex = 0
        maxvalue = text_result2[text_i][0]
        for index, value in enumerate(text_result2[text_i]):
            if value > maxvalue:
                maxindex = index
                maxvalue = value
        if text_label[text_i][maxindex] == 1:
            ACR2 += 1
    print('absolute correct rate (ACR):')
    print(str(100 * ACR2 / len(text_data2)) + '%')
    #mat.append(text_result2)
    #np.savetxt('mat{}.txt'.format(i+1), text_result2, delimiter=',')

    # text_label = t0[10000:12500].tolist()

    # for text_i in range(0, len(text_result)):
    #     print(text_result[text_i])
    # print(text_label)

    ACR = 0

    for text_i in range(0, len(text_result)):
        maxindex = 0
        maxvalue = text_result[text_i][0]
        for index, value in enumerate(text_result[text_i]):
            if value > maxvalue:
                maxindex = index
                maxvalue = value
        if text_label[text_i][maxindex] == 1:
            ACR += 1
    print('absolute correct rate (ACR):')
    print(str(100 * ACR / len(text_data)) + '%')
gongju = np.zeros((2000, 3))
ceshi = ceshi / (gongju + 5)
print("ceshi:================")
print(ceshi)
#np.savetxt('matrix.txt', ceshi, delimiter=',')
print("mat:==================")
print(mat)
BCR = 0

for text_i in range(0, len(text_result)):
    maxindex = 0
    maxvalue = text_result[text_i][0]
    for index, value in enumerate(text_result[text_i]):
        if value > maxvalue:
            maxindex = index
            maxvalue = value
    if text_label[text_i][maxindex] == 1:
        BCR += 1
    elif text_label[text_i][0] != 1 and maxindex != 0:
        BCR += 1
print('basic correct rate (BCR):')
print(str(100*BCR/len(text_data))+'%')