[code]自定义二分类数据集以及利用保存的模型输出评价指标

 1.保存成npy的自定义数据集

import os
import numpy as np
import random
import pickle

def read_file(file_path):
    file = open(file_path, 'r')
    lines = file.readlines()
    content = []
    for line in lines:
        line = line.replace("\t", " ")
        line_split = line.strip("\r\n\t").strip().split(" ")
        content.append(line_split)
    return content


def get_file_list(folder):
    file_list = []
    for file in os.listdir(folder):
        file_list.append(os.path.join(folder, file))
    return file_list


if __name__ == '__main__':
    all_file = get_file_list("文件夹路径")
    data = [(read_file(item)) for item in all_file]
    data = np.array(data)
    np.save('xxxxx.npy', data)
    print(x)

2.用于二分类的自定义数据集

import os
import numpy as np
import random
import pickle

def read_file(file_path):
    file = open(file_path, 'r')
    lines = file.readlines()
    content = []
    for line in lines:
        line = line.replace("\t", " ")
        line_split = line.strip("\r\n\t").strip().split(" ")
        content.append(line_split)
    return content


def get_file_list(folder):
    file_list = []
    for file in os.listdir(folder):
        file_list.append(os.path.join(folder, file))
    return file_list


def savepkl(data, path):
    file = open(path, 'wb')
    pickle.dump(data, file)
    file.close()


if __name__ == '__main__':
    FOLDERS = [
    {"class": 1, "folder": "分类为1的文件夹路径",
    {"class": 0, "folder": "分类为0的文件夹路径"
    ]
    all_file = [(item, folder["class"]) for folder in FOLDERS for item in get_file_list(folder["folder"])]
    random.shuffle(all_file)
    x = [(read_file((item[0]))) for item in all_file]
    y = [(item[1]) for item in all_file]
    x_np = np.array(x)
    y_np = np.array(y)
    # 保存数据和标签的pkl
    x_pklfile = "XXX.pkl" 
    y_pklfile = "XXX.pkl" 
    savepkl(x_np, x_pklfile)
    savepkl(y_np, y_pklfile)

3.利用保存的模型输出评价指标

from sklearn.metrics import confusion_matrix, f1_score, precision_score, recall_score, accuracy_score
import pickle
import numpy as np
from keras.models import load_model
from keras.utils import np_utils

if __name__ == '__main__':
    # 输入数据、标签和模型的地址
    pklfilex = "xxx.pkl"
    pklfiley = "xxx.pkl"
    model_name = 'xxx.h5
    # 导入数据、标签和模型
    np_all_samples_x = pickle.load(open(pklfilex, 'rb'))
    np_all_samples_y = pickle.load(open(pklfiley, 'rb'))
    np_all_samples_x = np.array(np_all_samples_x, dtype=int)
    file_num = len(np_all_samples_y)
    sub_file_num = int(file_num / FOLD)
    x_test = np_all_samples_x[0: sub_file_num]  # The samples for testing
    y_test = np_all_samples_y[0: sub_file_num]  # The label of the samples for testing
    # 模型预测
    model = load_model(model_name)
    y_pred = np.argmax(model.predict(x_test), axis=1)
    # 输出评价指标
    accuracy = accuracy_score(y_test, y_pred)
    precision = precision_score(y_test, y_pred)
    recall = recall_score(y_test, y_pred, average='binary')
    f1score = f1_score(y_test, y_pred, average='binary')
    MiF1 = f1_score(y_test, y_pred, average='micro')
    MaF1 = f1_score(y_test, y_pred, average='macro')
    print('accuracy:', accuracy, 'precision:', precision)
    print('recall:', recall, 'f1score:', f1score)
    print('Macro-F1: {}'.format(MaF1), 'Micro-F1: {}'.format(MiF1))