Multivariate Time Series Forecasting With Lstms In Keras 2d5m6a

Multivariate Time Series Forecasting with LSTMs in Keras machinelearningmastery.com/multivariate-time-series-forecasting-lstms-keras Jason Brownlee

August 13, 2017

Neural networks like Long Short-Term Memory (LSTM) recurrent neural networks are able to almost seamlessly model problems with multiple input variables. This is a great benefit in time series forecasting, where classical linear methods can be difficult to adapt to multivariate or multiple input forecasting problems. In this tutorial, you will discover how you can develop an LSTM model for multivariate time series forecasting in the Keras deep learning library. After completing this tutorial, you will know: How to transform a raw dataset into something we can use for time series forecasting. How to prepare data and fit an LSTM for a multivariate time series forecasting problem. How to make a forecast and rescale the result back into the original units. Let’s get started. Updated Aug/2017: Fixed a bug where yhat was compared to obs at the previous time step when calculating the final RMSE. Thanks, Songbin Xu and David Righart. Update Oct/2017: Added a new example showing how to train on multiple prior time steps due to popular demand.

Tutorial Overview This tutorial is divided into 3 parts; they are: 1. Air Pollution Forecasting 2. Basic Data Preparation 3. Multivariate LSTM Forecast Model

Python Environment This tutorial assumes you have a Python SciPy environment installed. You can use either Python 2 or 3 with this tutorial. You must have Keras (2.0 or higher) installed with either the TensorFlow or Theano backend. The tutorial also assumes you have scikit-learn, Pandas, NumPy and Matplotlib installed. If you need help with your environment, see this post:

How to Setup a Python Environment for Machine Learning and Deep Learning with Anaconda

Need help with LSTMs for Sequence Prediction? Take my free 7-day email course and discover 6 different LSTM architectures (with code). Click to and also get a free PDF Ebook version of the course. Start Your FREE Mini-Course Now!

1. Air Pollution Forecasting In this tutorial, we are going to use the Air Quality dataset. This is a dataset that reports on the weather and the level of pollution each hour for five years at the US embassy in Beijing, China. The data includes the date-time, the pollution called PM2.5 concentration, and the weather information including dew point, temperature, pressure, wind direction, wind speed and the cumulative number of hours of snow and rain. The complete feature list in the raw data is as follows: 1. No: row number 2. year: year of data in this row 3. month: month of data in this row 4. day: day of data in this row 5. hour: hour of data in this row 6. pm2.5: PM2.5 concentration 7. DEWP: Dew Point 8. TEMP: Temperature 9. PRES: Pressure 10. cbwd: Combined wind direction 11. Iws: Cumulated wind speed 12. Is: Cumulated hours of snow 13. Ir: Cumulated hours of rain We can use this data and frame a forecasting problem where, given the weather conditions and pollution for prior hours, we forecast the pollution at the next hour. This dataset can be used to frame other forecasting problems. Do you have good ideas? Let me know in the comments below. You can the dataset from the UCI Machine Learning Repository. Beijing PM2.5 Data Set the dataset and place it in your current working directory with the filename

“raw.csv“.

2. Basic Data Preparation The data is not ready to use. We must prepare it first. Below are the first few rows of the raw dataset. 1

No,year,month,day,hour,pm2.5,DEWP,TEMP,PRES,cbwd,Iws,Is,Ir

2 3 4

1,2010,1,1,0,NA,-21,-11,1021,NW,1.79,0,0

5 6

4,2010,1,1,3,NA,-21,-14,1019,NW,9.84,0,0 5,2010,1,1,4,NA,-20,-12,1018,NW,12.97,0,0

2,2010,1,1,1,NA,-21,-12,1020,NW,4.92,0,0 3,2010,1,1,2,NA,-21,-11,1019,NW,6.71,0,0

The first step is to consolidate the date-time information into a single date-time so that we can use it as an index in Pandas. A quick check reveals NA values for pm2.5 for the first 24 hours. We will, therefore, need to remove the first row of data. There are also a few scattered “NA” values later in the dataset; we can mark them with 0 values for now. The script below loads the raw dataset and parses the date-time information as the Pandas DataFrame index. The “No” column is dropped and then clearer names are specified for each column. Finally, the NA values are replaced with “0” values and the first 24 hours are removed. The “No” column is dropped and then clearer names are specified for each column. Finally, the NA values are replaced with “0” values and the first 24 hours are removed.

1 2 3

from pandas import read_csv from datetime import datetime # load data

4 5 6

def parse(x): returndatetime.strptime(x,'%Y %m %d %H') dataset=read_csv('raw.csv',parse_dates=

7 8

[['year','month','day','hour']],index_col=0,date_parser=parse) dataset.drop('No',axis=1,inplace=True) # manually specify column names

9 10 11 12

dataset.columns=['pollution','dew','temp','press','wnd_dir','wnd_spd','snow','rain'] dataset.index.name='date'

13

dataset['pollution'].fillna(0,inplace=True)

14

# drop the first 24 hours

15 16

dataset=dataset[24:]

17

print(dataset.head(5)) # save to file

18

# mark all NA values with 0

# summarize first 5 rows

Running the example prints the first 5 rows of the transformed dataset and saves the dataset to “pollution.csv“. 1 2 3

                     pollution  dew  temp   press wnd_dir  wnd_spd  snow  rain date 2010-01-02 00:00:00      129.0  -16  -4.0  1020.0      SE     1.79     0     0

4 5 6

2010-01-02 01:00:00      148.0  -15  -4.0  1020.0      SE     2.68     0     0 2010-01-02 02:00:00      159.0  -11  -5.0  1021.0      SE     3.57     0     0 2010-01-02 03:00:00      181.0   -7  -5.0  1022.0      SE     5.36     1     0

7

2010-01-02 04:00:00      138.0   -7  -5.0  1022.0      SE     6.25     2     0

Now that we have the data in an easy-to-use form, we can create a quick plot of each series and see what we have. The code below loads the new “pollution.csv” file and plots each series as a separate subplot, except wind speed dir, which is categorical.

1 2 3

from pandas import read_csv from matplotlib import pyplot # load dataset

4 5 6

dataset=read_csv('pollution.csv',header=0,index_col=0) values=dataset.values # specify columns to plot

7 8

groups=[0,1,2,3,5,6,7] i=1

9

# plot each column

10 11

pyplot.figure() forgroup ingroups:

Running the example creates a plot with 7 subplots showing the 5 years of data for each variable.

Line Plots of Air Pollution Time Series

3. Multivariate LSTM Forecast Model In this section, we will fit an LSTM to the problem.

LSTM Data Preparation The first step is to prepare the pollution dataset for the LSTM. This involves framing the dataset as a supervised learning problem and normalizing the input variables. We will frame the supervised learning problem as predicting the pollution at the current hour (t) given the pollution measurement and weather conditions at the prior time step. This formulation is straightforward and just for this demonstration. Some alternate formulations you could explore include: Predict the pollution for the next hour based on the weather conditions and pollution over the last 24 hours. Predict the pollution for the next hour as above and given the “expected” weather conditions for the next hour. We can transform the dataset using the series_to_supervised() function developed in the blog post: How to Convert a Time Series to a Supervised Learning Problem in Python First, the “pollution.csv” dataset is loaded. The wind speed feature is label encoded (integer encoded). This could further be one-hot encoded in the future if you are interested in exploring it. Next, all features are normalized, then the dataset is transformed into a supervised learning problem. The weather variables for the hour to be predicted (t) are then removed. The complete code listing is provided below.

1 2 3

# convert series to supervised learning def series_to_supervised(data,n_in=1,n_out=1,dropnan=True): n_vars=1iftype(data)islist elsedata.shape[1]

4 5 6

df=DataFrame(data) cols,names=list(),list() # input sequence (t-n, ... t-1)

7 8

foriinrange(n_in,0,-1): cols.append(df.shift(i)) names+=[('var%d(t-%d)'%(j+1,i))forjinrange(n_vars)]

9 10 11 12

# forecast sequence (t, t+1, ... t+n) foriinrange(0,n_out):

13

ifi==0:

14

names+=[('var%d(t)'%(j+1))forjinrange(n_vars)]

15 16

else:

17

# put it all together

cols.append(df.shift(-i))

names+=[('var%d(t+%d)'%(j+1,i))forjinrange(n_vars)]

Running the example prints the first 5 rows of the transformed dataset. We can see the 8 input variables (input series) and the 1 output variable (pollution level at the current hour).

1 2 3

   var1(t-1)  var2(t-1)  var3(t-1)  var4(t-1)  var5(t-1)  var6(t-1)  \ 1   0.129779   0.352941   0.245902   0.527273   0.666667   0.002290 2   0.148893   0.367647   0.245902   0.527273   0.666667   0.003811

4 5

3   0.159960   0.426471   0.229508   0.545454   0.666667   0.005332 4   0.182093   0.485294   0.229508   0.563637   0.666667   0.008391

This data preparation is simple and there is more we could explore. Some ideas you could look at include: One-hot encoding wind speed. Making all series stationary with differencing and seasonal adjustment. Providing more than 1 hour of input time steps. This last point is perhaps the most important given the use of Backpropagation through time by LSTMs when learning sequence prediction problems.

Define and Fit Model In this section, we will fit an LSTM on the multivariate input data. First, we must split the prepared dataset into train and test sets. To speed up the training of the model for this demonstration, we will only fit the model on the first year of data, then evaluate it on the remaining 4 years of data. If you have time, consider exploring the inverted version of this test harness. The example below splits the dataset into train and test sets, then splits the train and test sets into input and output variables. Finally, the inputs (X) are reshaped into the 3D format expected by LSTMs, namely [samples, timesteps, features].

1 2 3

# split into train and test sets values=reframed.values n_train_hours=365*24

4 5 6

train=values[:n_train_hours,:] test=values[n_train_hours:,:] # split into input and outputs

7 8

train_X,train_y=train[:,:-1],train[:,-1] test_X,test_y=test[:,:-1],test[:,-1]

9

# reshape input to be 3D [samples, timesteps, features]

10 11

train_X=train_X.reshape((train_X.shape[0],1,train_X.shape[1])) test_X=test_X.reshape((test_X.shape[0],1,test_X.shape[1]))

Running this example prints the shape of the train and test input and output sets with about 9K hours of data for training and about 35K hours for testing. 1

(8760, 1, 8) (8760,) (35039, 1, 8) (35039,)

Now we can define and fit our LSTM model. We will define the LSTM with 50 neurons in the first hidden layer and 1 neuron in the output layer for predicting pollution. The input shape will be 1 time step with 8 features. We will use the Mean Absolute Error (MAE) loss function and the efficient Adam version of stochastic gradient descent. The model will be fit for 50 training epochs with a batch size of 72. that the internal state of the LSTM in Keras is reset at the end of each batch, so an internal state that is a function of a number of days may be helpful (try testing this). Finally, we keep track of both the training and test loss during training by setting the validation_data argument in the fit() function. At the end of the run both the training and test loss are plotted. 1 2

# design network model=Sequential()

3 4

model.add(LSTM(50,input_shape=(train_X.shape[1],train_X.shape[2]))) model.add(Dense(1))

5 6

model.compile(loss='mae',optimizer='adam') # fit network

7

history=model.fit(train_X,train_y,epochs=50,batch_size=72,validation_data=

8 9

(test_X,test_y),verbose=2,shuffle=False) # plot history

10 11 12

pyplot.plot(history.history['loss'],label='train') pyplot.plot(history.history['val_loss'],label='test') pyplot.legend() pyplot.show()

Evaluate Model After the model is fit, we can forecast for the entire test dataset. We combine the forecast with the test dataset and invert the scaling. We also invert scaling on the test dataset with the expected pollution numbers. With forecasts and actual values in their original scale, we can then calculate an error score for the model. In this case, we calculate the Root Mean Squared Error (RMSE) that gives error in the same units as the variable itself. 1 2 3

# make a prediction yhat=model.predict(test_X) test_X=test_X.reshape((test_X.shape[0],test_X.shape[2]))

4

# invert scaling for forecast

5 6 7

inv_yhat=concatenate((yhat,test_X[:,1:]),axis=1) inv_yhat=scaler.inverse_transform(inv_yhat) inv_yhat=inv_yhat[:,0]

8 9 10

# invert scaling for actual test_y=test_y.reshape((len(test_y),1)) inv_y=concatenate((test_y,test_X[:,1:]),axis=1)

11 12 13

inv_y=scaler.inverse_transform(inv_y) inv_y=inv_y[:,0] # calculate RMSE

14 15

rmse=sqrt(mean_squared_error(inv_y,inv_yhat)) print('Test RMSE: %.3f'%rmse)

Complete Example The complete example is listed below. NOTE: This example assumes you have prepared the data correctly, e.g. converted the ed “raw.csv” to the prepared “pollution.csv“. See the first part of this tutorial.

1 2 3

from math import sqrt from numpy import concatenate from matplotlib import pyplot

4 5 6

from pandas import read_csv from pandas import DataFrame from pandas import concat

7 8

from sklearn.preprocessing import MinMaxScaler from sklearn.preprocessing import LabelEncoder

9

from sklearn.metrics import mean_squared_error

10 11 12

from keras.models import Sequential from keras.layers import Dense from keras.layers import LSTM

13

# convert series to supervised learning

14

def series_to_supervised(data,n_in=1,n_out=1,dropnan=True):

15

n_vars=1iftype(data)islist elsedata.shape[1]

16 17

df=DataFrame(data) cols,names=list(),list()

18 19 20

# input sequence (t-n, ... t-1) foriinrange(n_in,0,-1): cols.append(df.shift(i))

21 22 23 24

names+=[('var%d(t-%d)'%(j+1,i))forjinrange(n_vars)] # forecast sequence (t, t+1, ... t+n) foriinrange(0,n_out): cols.append(df.shift(-i))

25 26 27

ifi==0: names+=[('var%d(t)'%(j+1))forjinrange(n_vars)] else:

28 29 30

names+=[('var%d(t+%d)'%(j+1,i))forjinrange(n_vars)] # put it all together agg=concat(cols,axis=1)

31 32 33

agg.columns=names # drop rows with NaN values ifdropnan:

34 35 36

agg.dropna(inplace=True) returnagg # load dataset

37 38 39 40

dataset=read_csv('pollution.csv',header=0,index_col=0) values=dataset.values # integer encode direction

41

values[:,4]=encoder.fit_transform(values[:,4]) # ensure all data is float

42 43

encoder=LabelEncoder()

values=values.astype('float32')

44

# normalize features

45 46

scaler=MinMaxScaler(feature_range=(0,1)) scaled=scaler.fit_transform(values)

47 48 49

# frame as supervised learning reframed=series_to_supervised(scaled,1,1) # drop columns we don't want to predict

50 51 52

reframed.drop(reframed.columns[[9,10,11,12,13,14,15]],axis=1,inplace=True) print(reframed.head()) # split into train and test sets

53 54 55

values=reframed.values n_train_hours=365*24 train=values[:n_train_hours,:]

56 57

test=values[n_train_hours:,:] # split into input and outputs

58

train_X,train_y=train[:,:-1],train[:,-1]

59 60 61

test_X,test_y=test[:,:-1],test[:,-1] # reshape input to be 3D [samples, timesteps, features] train_X=train_X.reshape((train_X.shape[0],1,train_X.shape[1]))

62

test_X=test_X.reshape((test_X.shape[0],1,test_X.shape[1]))

63

print(train_X.shape,train_y.shape,test_X.shape,test_y.shape)

64 65

# design network model=Sequential()

66

model.add(LSTM(50,input_shape=(train_X.shape[1],train_X.shape[2])))

67 68

model.add(Dense(1)) model.compile(loss='mae',optimizer='adam')

69 70 71

# fit network history=model.fit(train_X,train_y,epochs=50,batch_size=72,validation_data= (test_X,test_y),verbose=2,shuffle=False)

72 73 74

# plot history pyplot.plot(history.history['loss'],label='train') pyplot.plot(history.history['val_loss'],label='test')

75 76 77 78

pyplot.legend() pyplot.show() # make a prediction yhat=model.predict(test_X)

79 80 81

test_X=test_X.reshape((test_X.shape[0],test_X.shape[2])) # invert scaling for forecast inv_yhat=concatenate((yhat,test_X[:,1:]),axis=1)

82 83 84

inv_yhat=scaler.inverse_transform(inv_yhat) inv_yhat=inv_yhat[:,0] # invert scaling for actual

85 86 87

test_y=test_y.reshape((len(test_y),1)) inv_y=concatenate((test_y,test_X[:,1:]),axis=1) inv_y=scaler.inverse_transform(inv_y)

88 89 90

inv_y=inv_y[:,0] # calculate RMSE rmse=sqrt(mean_squared_error(inv_y,inv_yhat))

91 92 93

print('Test RMSE: %.3f'%rmse)

94 95

Running the example first creates a plot showing the train and test loss during training. Interestingly, we can see that test loss drops below training loss. The model may be

overfitting the training data. Measuring and plotting RMSE during training may shed more light on this.

Line Plot of Train and Test Loss from the Multivariate LSTM During Training

The Train and test loss are printed at the end of each training epoch. At the end of the run, the final RMSE of the model on the test dataset is printed. We can see that the model achieves a respectable RMSE of 26.496, which is lower than an RMSE of 30 found with a persistence model. 1 2 3

... Epoch 46/50 0s - loss: 0.0143 - val_loss: 0.0133

4 5 6

Epoch 47/50

7

0s - loss: 0.0144 - val_loss: 0.0133

8 9

Epoch 49/50 0s - loss: 0.0143 - val_loss: 0.0133

10

Epoch 50/50

11 12

0s - loss: 0.0144 - val_loss: 0.0133 Test RMSE: 26.496

0s - loss: 0.0143 - val_loss: 0.0133 Epoch 48/50

This model is not tuned. Can you do better? Let me know your problem framing, model configuration, and RMSE in the comments

below.

Update: Train On Multiple Lag Timesteps Example There have been many requests for advice on how to adapt the above example to train the model on multiple previous time steps. I had tried this and a myriad of other configurations when writing the original post and decided not to include them because they did not lift model skill. Nevertheless, I have included this example below as reference template that you could adapt for your own problems. The changes needed to train the model on multiple previous time steps are quite minimal, as follows: First, you must frame the problem suitably when calling series_to_supervised(). We will use 3 hours of data as input. Also note, we no longer explictly drop the columns from all of the other fields at ob(t). 1 2

# specify the number of lag hours n_hours=3

3 4 5

n_features=8 # frame as supervised learning reframed=series_to_supervised(scaled,n_hours,1)

Next, we need to be more careful in specifying the column for input and output. We have 3 * 8 + 8 columns in our framed dataset. We will take 3 * 8 or 24 columns as input for the obs of all features across the previous 3 hours. We will take just the pollution variable as output at the following hour, as follows: 1

# split into input and outputs

2

n_obs=n_hours *n_features train_X,train_y=train[:,:n_obs],train[:,-n_features]

3 4 5

test_X,test_y=test[:,:n_obs],test[:,-n_features] print(train_X.shape,len(train_X),train_y.shape)

Next, we can reshape our input data correctly to reflect the time steps and features. 1

# reshape input to be 3D [samples, timesteps, features]

2 3

train_X=train_X.reshape((train_X.shape[0],n_hours,n_features)) test_X=test_X.reshape((test_X.shape[0],n_hours,n_features))

Fitting the model is the same. The only other small change is in how to evaluate the model. Specifically, in how we reconstruct the rows with 8 columns suitable for reversing the scaling operation to get the y

and yhat back into the original scale so that we can calculate the RMSE. The gist of the change is that we concatenate the y or yhat column with the last 7 features of the test dataset in order to inverse the scaling, as follows: 1 2 3 4

# invert scaling for forecast inv_yhat=concatenate((yhat,test_X[:,-7:]),axis=1)

5 6 7

# invert scaling for actual test_y=test_y.reshape((len(test_y),1))

8 9

inv_y=scaler.inverse_transform(inv_y) inv_y=inv_y[:,0]

inv_yhat=scaler.inverse_transform(inv_yhat) inv_yhat=inv_yhat[:,0]

inv_y=concatenate((test_y,test_X[:,-7:]),axis=1)

We can tie all of these modifications to the above example together. The complete example of multvariate time series forecasting with multiple lag inputs is listed below:

1 2 3

from math import sqrt from numpy import concatenate from matplotlib import pyplot

4 5 6

from pandas import read_csv from pandas import DataFrame from pandas import concat

7 8

from sklearn.preprocessing import MinMaxScaler from sklearn.preprocessing import LabelEncoder

9

from sklearn.metrics import mean_squared_error

10 11 12

from keras.models import Sequential from keras.layers import Dense from keras.layers import LSTM

13

# convert series to supervised learning

14

def series_to_supervised(data,n_in=1,n_out=1,dropnan=True):

15 16

n_vars=1iftype(data)islist elsedata.shape[1] df=DataFrame(data)

17

cols,names=list(),list()

18 19

# input sequence (t-n, ... t-1) foriinrange(n_in,0,-1):

20 21 22

cols.append(df.shift(i)) names+=[('var%d(t-%d)'%(j+1,i))forjinrange(n_vars)] # forecast sequence (t, t+1, ... t+n)

23 24 25

foriinrange(0,n_out): cols.append(df.shift(-i)) ifi==0:

26 27 28 29

names+=[('var%d(t)'%(j+1))forjinrange(n_vars)] else: names+=[('var%d(t+%d)'%(j+1,i))forjinrange(n_vars)] # put it all together

30 31

agg=concat(cols,axis=1) agg.columns=names

32

# drop rows with NaN values

33 34

ifdropnan:

35 36

agg.dropna(inplace=True) returnagg # load dataset

37 38 39

dataset=read_csv('pollution.csv',header=0,index_col=0) values=dataset.values # integer encode direction

40 41 42

encoder=LabelEncoder() values[:,4]=encoder.fit_transform(values[:,4]) # ensure all data is float

43 44 45

values=values.astype('float32') # normalize features scaler=MinMaxScaler(feature_range=(0,1))

46 47

scaled=scaler.fit_transform(values) # specify the number of lag hours

48

n_hours=3

49

n_features=8

50 51 52

# frame as supervised learning reframed=series_to_supervised(scaled,n_hours,1) print(reframed.shape)

53 54 55

# split into train and test sets values=reframed.values n_train_hours=365*24

56 57

train=values[:n_train_hours,:] test=values[n_train_hours:,:]

58

# split into input and outputs

59 60 61

n_obs=n_hours *n_features train_X,train_y=train[:,:n_obs],train[:,-n_features] test_X,test_y=test[:,:n_obs],test[:,-n_features]

62

print(train_X.shape,len(train_X),train_y.shape)

63

# reshape input to be 3D [samples, timesteps, features]

64 65

train_X=train_X.reshape((train_X.shape[0],n_hours,n_features)) test_X=test_X.reshape((test_X.shape[0],n_hours,n_features))

66

print(train_X.shape,train_y.shape,test_X.shape,test_y.shape)

67 68

# design network model=Sequential()

69 70 71

model.add(LSTM(50,input_shape=(train_X.shape[1],train_X.shape[2]))) model.add(Dense(1)) model.compile(loss='mae',optimizer='adam')

72 73 74

# fit network history=model.fit(train_X,train_y,epochs=50,batch_size=72,validation_data= (test_X,test_y),verbose=2,shuffle=False)

75 76 77 78

# plot history pyplot.plot(history.history['loss'],label='train') pyplot.plot(history.history['val_loss'],label='test') pyplot.legend()

79 80 81

pyplot.show() # make a prediction yhat=model.predict(test_X)

82 83 84

test_X=test_X.reshape((test_X.shape[0],n_hours*n_features)) # invert scaling for forecast inv_yhat=concatenate((yhat,test_X[:,-7:]),axis=1)

85 86 87

inv_yhat=scaler.inverse_transform(inv_yhat) inv_yhat=inv_yhat[:,0] # invert scaling for actual

88 89 90

test_y=test_y.reshape((len(test_y),1)) inv_y=concatenate((test_y,test_X[:,-7:]),axis=1) inv_y=scaler.inverse_transform(inv_y)

91 92 93

inv_y=inv_y[:,0] # calculate RMSE rmse=sqrt(mean_squared_error(inv_y,inv_yhat))

94

print('Test RMSE: %.3f'%rmse)

95 96 97 98

The model is fit as before in a minute or two. 1 2

... Epoch 45/50

3

1s - loss: 0.0143 - val_loss: 0.0154

4 5

Epoch 46/50 1s - loss: 0.0143 - val_loss: 0.0148

6

Epoch 47/50

7

1s - loss: 0.0143 - val_loss: 0.0152

8 9

Epoch 48/50 1s - loss: 0.0143 - val_loss: 0.0151

10

Epoch 49/50

11 12

1s - loss: 0.0143 - val_loss: 0.0152 Epoch 50/50

13

1s - loss: 0.0144 - val_loss: 0.0149

A plot of train and test loss over the epochs is plotted.

Plot of Loss on the Train and Test Datasets

Finally, the Test RMSE is printed, not really showing any advantage in skill, at least on this problem. 1

Test RMSE: 27.177

I would add that the LSTM does not appear to be suitable for autoregression type problems

and that you may be better off exploring an MLP with a large window. I hope this example helps you with your own time series forecasting experiments.

Further Reading This section provides more resources on the topic if you are looking go deeper. Beijing PM2.5 Data Set on the UCI Machine Learning Repository The 5 Step Life-Cycle for Long Short-Term Memory Models in Keras Time Series Forecasting with the Long Short-Term Memory Network in Python Multi-step Time Series Forecasting with Long Short-Term Memory Networks in Python

Summary In this tutorial, you discovered how to fit an LSTM to a multivariate time series forecasting problem. Specifically, you learned: How to transform a raw dataset into something we can use for time series forecasting. How to prepare data and fit an LSTM for a multivariate time series forecasting problem. How to make a forecast and rescale the result back into the original units. Do you have any questions? Ask your questions in the comments below and I will do my best to answer.

Develop LSTMs for Sequence Prediction Today!

Develop Your Own LSTM models in Minutes …with just a few lines of python code Discover how in my new Ebook: Long Short-Term Memory Networks with Python It provides self-study tutorials on topics like: CNN LSTMs, Encoder-Decoder LSTMs, generative models, data preparation, making predictions and much more…

Finally Bring LSTM Recurrent Neural Networks to Your Sequence Predictions Projects Skip the Academics. Just Results. Click to learn more.