read_path_val = "./data/test" file_path_list = traversalDir_FirstDir(read_path_val) test_data = merge(file_path_list)[diying_attribute] read_path_val = "./data/test_set" file_path_list = traversalDir_FirstDir(read_path_val) sources = np.zeros((142276, 24)) for i in range(len(file_path_list)): data = pd.read_csv(file_path_list[i], engine="python")["Class"] sources[:, i] = data.values sources = np.sum(sources, axis=1) source_label = [1 if source > 0 else 0 for source in sources] test_data["Class"] = source_label # 构建训练集和测试集 train = pd.concat([train_data, val_data], axis=0) train.iloc[:, 0:-1] = z_norm(train.iloc[:, 0:-1].copy()) train = train.values np.random.shuffle(train) train_x = train[:, 0:-1] train_y = train[:, -1] test = test_data test.iloc[:, 0:-1] = z_norm(test.iloc[:, 0:-1].copy()) test = test.values np.random.shuffle(test) test_x = test[:, 0:-1] test_y = test[:, -1] # 打印输出训练集和测试集的信息 print("--------------------") print("训练集样本大小为:", train_x.shape[0]) print("训练集正常样本大小为:", train_x.shape[0] - np.sum(train_y))
config = tf.ConfigProto() config.gpu_options.allow_growth = True # 不全部占满显存, 按需分配 sess = tf.Session(config=config) KTF.set_session(sess) # 训练模型 read_path = "./data/train" read_file_list = traversalDir_FirstDir(read_path) data = merge(read_file_list) diying_attribute = list( pd.read_csv("./parameter/diying.csv", header=None)[0]) for i in range(len(diying_attribute)): diying_attribute[i] = "BX0101_" + diying_attribute[i] data = data[diying_attribute] # 数据标准化 data = z_norm(data) for i in range(0, 6): # 创建模型 model = build_model(sequence_length) # 得到训练数据 tmp = data.iloc[:, i] train_x, train_y = GetData(tmp.values, sequence_length) print(str(i) + data.columns[i] + " start!" + "Training...") print(train_x.shape) print(train_y.shape) history = LossHistory() model.fit(train_x, train_y, batch_size=batch_size,