python数据分析实例代码 - 家用电器用户行为分析与事件识别.rar

import pandas as pd import numpy as np from keras.layers.core import Activation, Dense from keras.models import Sequential ''' session_1() 数据筛选 session_2() 阈值寻优 session_3() 建立神经网络模型，并对模型进行检验 session_4() 根据特征推算是否满足某项条件 ''' def session_1(): # 阈值 threshold = pd.Timedelta('4 min') data = pd.read_excel('water_heater.xls') # DataFrame处理 data['发生时间'] = pd.to_datetime(data['发生时间'], format='%Y%m%d%H%M%S') data = data[data['水流量'] > 0] # 只记录流量大于0的数据 d = data['发生时间'].diff() > threshold # 相邻时间做差分，大于threshold data['事件编号'] = d.cumsum() + 1 # 通过累计求和的方式为事件编号 data.to_csv('divid_sequence.csv') def session_2(): # 使用之后四个点的平均斜率 n = 4 # 专家阈值 threshold = pd.Timedelta(minutes=5) data = pd.read_excel('water_heater.xls') data['发生时间'] = pd.to_datetime(data['发生时间'], format='%Y%m%d%H%M%S') data = data[data['水流量'] > 0] # 定义阈值列 dt = [pd.Timedelta(minutes=i) for i in np.arange(1, 9, 0.25)] h = pd.DataFrame(dt, columns=['阈值']) # 相邻时间的差分，比较是否大于阈值 def event_num(ts): d = data['发生时间'].diff() > ts # 返回事件数 return d.sum() + 1 # 计算每个阈值对应的事件数 h['事件数'] = h['阈值'].apply(event_num) # 计算每两个相邻点对应的斜率 h['斜率'] = h['事件数'].diff() / 0.25 # 采用后n个斜率的绝对值的平均作为斜率指标 h['斜率指标'] = pd.Series.rolling(h['斜率'].abs(), n).mean() ts = h['阈值'][h['斜率指标'].idxmin() - n] if ts > threshold: ts = pd.Timedelta(minutes=4) print(ts) def session_3(): # 读取训练集和测试集，并且划分样本特征和标签 data_train = pd.read_excel('train_neural_network_data.xls') data_test = pd.read_excel('test_neural_network_data.xls') y_train = data_train.iloc[:, 4].values x_train = data_train.iloc[:, 5:17].values y_test = data_test.iloc[:, 4].values x_test = data_test.iloc[:, 5:17].values # 建立神经网络模型 model = Sequential() model.add(Dense(17, input_shape=(11,))) model.add(Activation('relu')) model.add(Dense(10, input_shape=(17,))) model.add(Activation('relu')) model.add(Dense(1, input_shape=(10,))) model.add(Activation('sigmoid')) # 编译模型 model.compile( loss='binary_crossentropy', optimizer='adam', sample_weight_mode='binary' ) # 训练模型 model.fit(x_train, y_train, epochs=100, batch_size=1) # 保存模型 model.save_weights('net.model') # 进行预测 r = pd.DataFrame(model.predict_classes(x_test), columns=['预测结果']) pd.concat([data_test.iloc[:, :5], r], axis=1).to_csv('test_output_data.csv') model.predict(x_test) return y_test def session_4(): threshold = pd.Timedelta('4 min') data = pd.read_excel('water_heater.xls') data['发生时间'] = pd.to_datetime(data['发生时间'], format='%Y%m%d%H%M%S') data = data[data['水流量'] > 0] d = data['发生时间'].diff() > threshold data['事件编号'] = d.cumsum() + 1 data_g = data.groupby('事件编号') result = pd.DataFrame() dt = pd.Timedelta(seconds=2) for _, g in data_g: temp = pd.DataFrame(index=[0]) # 根据用水时长、开关机切换次数、总用水量退出是否洗澡 t_start = g['发生时间'].min() t_end = g['发生时间'].max() temp['用水事件时长(M)'] = (dt+t_end-t_start).total_seconds()/60 temp['开关机切换次数'] = (pd.Series.rolling(g['开关机状态'] == '关',2).sum() == 1).sum() temp['总用水量(L)'] = g['水流量'].sum() t_diff = g['发生时间'].diff() if len(g['发生时间']) == 1: temp["总用水时长（Min）"] = dt.total_seconds() / 60 else: temp["总用水时长（Min）"] = ( t_diff.sum() - t_diff.iloc[1] / 2 - t_diff.iloc[len(t_diff) - 1] / 2).total_seconds() / 60 temp["平均水流量（L/min）"] = temp["总用水量(L)"] / temp["总用水时长（Min）"] result = result.append(temp, ignore_index=True) result.to_csv('attribute_extract.csv') if __name__ == '__main__': # session_1() # session_2() # session_3() session_4()