天气预测.rar天气预测.rar天气预测.rar

# 基于 keras LSTM的天气预测 时间序列预测 完整代码+数据 import pandas as pd import numpy as np # from tensorflow import keras from keras import layers from keras import Input from keras.models import Model from keras.models import Sequential from keras.layers.normalization import BatchNormalization from keras import callbacks from keras.optimizers import adam from keras.optimizers import RMSprop callback_list=[ callbacks.EarlyStopping(monitor="val_loss",patience=500), callbacks.ModelCheckpoint(filepath="weather_v2.h5",monitor="val_loss",save_best_only=True), callbacks.ReduceLROnPlateau(monitor="val_loss",factor=0.9,verbose=1,patience=10) ] a_weather=pd.read_csv("weather.csv") a_train=pd.read_csv("train_v2.csv") a_train=a_train.drop(['Unnamed: 0','Unnamed: 0.1'],axis=1) del a_weather["Unnamed: 0"] a_train=np.array(a_train) a_weather=np.array(a_weather) window=4 samples=np.zeros((a_train.shape[0]-window,window,a_train.shape[1])) targets=np.zeros((a_train.shape[0]-window,a_train.shape[1])) weather_targets=np.zeros((a_train.shape[0]-window,)) xr=0 zt=4 for i in range(samples.shape[0]): samples[i]=a_train[xr:xr+4] targets[i]=a_weather[zt] xr=xr+1 if xr+1<=samples.shape[0]-1 else samples.shape[0]-1 zt = zt + 1 if xr + 1 <= samples.shape[0] - 1 else samples.shape[0] - 1 samples=samples[:,:,1:] didital_targets=targets[:,2:7] weather_targets=targets[:,7:] print(samples.shape) print(didital_targets.shape) print(weather_targets.shape) weather_targets_onehot=np.zeros((weather_targets.shape[0],7)) for i in range(weather_targets.shape[0]): weather_targets_onehot[i][int(weather_targets[i])-1]=1 data_input=Input(shape=(samples.shape[1],samples.shape[-1])) x=layers.LSTM(128,return_sequences=True,activation="relu")(data_input) x=BatchNormalization()(x) x=layers.LSTM(256,return_sequences=False,activation="relu")(x) x=BatchNormalization()(x) x=layers.Dense(256,activation="relu")(x) x=BatchNormalization()(x) x=layers.Dense(512,activation="tanh")(x) x=BatchNormalization()(x) digital_predict=layers.Dense(5,name="digital")(x) weather_predict=layers.Dense(7,activation="softmax",name="weather")(x) weather_model=Model(data_input,[digital_predict,weather_predict]) print(weather_model.summary()) weather_model.compile(optimizer=adam(),loss=["mae","categorical_crossentropy"],loss_weights=[5,1],metrics=["accuracy"]) weather_model.fit(samples,[didital_targets,weather_targets_onehot],epochs=100,batch_size=30,callbacks=callback_list,validation_split=0.3)