LSTM Multi-state forecast












0












$begingroup$


I follow several tutorials about LSTM multi-step forecast to solve my problem.



My problem:
I have a time series of price about 3 months (Jan 2018 to Mar 2018) and I assume there are seasonal. So, my time series behaviors are changed because of such seasonal effects.



Q1
Do I need to apply sliding window to prevent learning from too long sequence. For example, each training sample takes prices of previous 14 days to predict the price of next 14 days. In other words, my input is (t-1), (t-2), ... , (t-7) and my output is (t), (t+1), ... , (t+6).



Q2
I read this post and feel that it's interesting by just passing a whole sequence and let the model makes decision by itself using the advantage of setting return_sequence=True and stateful=True. However, I'm a newbie to LSTM so I could not have clear understand how to apply it to the model according to such post.



Suppose my data look like this



import pandas as pd
import numpy as pd

np.random.seed(40)
df = pd.DataFrame({"Price":list(np.random.choice(range(1000, 1500), 14)) +
list(np.random.choice(range(1500, 2000), 14)) +
list(np.random.choice(range(1800, 2500), 14)) +
list(np.random.choice(range(1200, 1500), 14)) +
list(np.random.choice(range(1500, 2000), 14)) +
list(np.random.choice(range(1800, 2300), 14)) +
list(np.random.choice(range(2500, 3000), 14)) +
list(np.random.choice(range(2400, 2800), 14)) +
list(np.random.choice(range(2600, 3000), 14)) +
list(np.random.choice(range(2300, 2700), 14))},
index=pd.date_range(start="2018-03-25", periods=140))

df["Price"] = df["Price"].astype(float)
scaler = MinMaxScaler(feature_range=(0, 1))
df["Scaled"] = scaler.fit_transform(df["Price"].values.reshape(-1, 1))


If I configure my model as follows



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

model = Sequential()
model.add(LSTM(no_neuron, input_shape=(no_sample, timestep, no_feature)))
model.add(Dense(1))
model.compile(loss='mean_squared_error', optimizer='adam')


Is that make sense to set my timestep=40, no_sample=140, no_feature=1 for my problem? and How to define batch size?



Q3



In my settings, I really need to predict the price from Apr 2018 until Dec 2018 is that possible to do even I have only data from Jan 2018 to Mar 2018?









share







New contributor




Ratchainant Thammasudjarit is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.







$endgroup$

















    0












    $begingroup$


    I follow several tutorials about LSTM multi-step forecast to solve my problem.



    My problem:
    I have a time series of price about 3 months (Jan 2018 to Mar 2018) and I assume there are seasonal. So, my time series behaviors are changed because of such seasonal effects.



    Q1
    Do I need to apply sliding window to prevent learning from too long sequence. For example, each training sample takes prices of previous 14 days to predict the price of next 14 days. In other words, my input is (t-1), (t-2), ... , (t-7) and my output is (t), (t+1), ... , (t+6).



    Q2
    I read this post and feel that it's interesting by just passing a whole sequence and let the model makes decision by itself using the advantage of setting return_sequence=True and stateful=True. However, I'm a newbie to LSTM so I could not have clear understand how to apply it to the model according to such post.



    Suppose my data look like this



    import pandas as pd
    import numpy as pd

    np.random.seed(40)
    df = pd.DataFrame({"Price":list(np.random.choice(range(1000, 1500), 14)) +
    list(np.random.choice(range(1500, 2000), 14)) +
    list(np.random.choice(range(1800, 2500), 14)) +
    list(np.random.choice(range(1200, 1500), 14)) +
    list(np.random.choice(range(1500, 2000), 14)) +
    list(np.random.choice(range(1800, 2300), 14)) +
    list(np.random.choice(range(2500, 3000), 14)) +
    list(np.random.choice(range(2400, 2800), 14)) +
    list(np.random.choice(range(2600, 3000), 14)) +
    list(np.random.choice(range(2300, 2700), 14))},
    index=pd.date_range(start="2018-03-25", periods=140))

    df["Price"] = df["Price"].astype(float)
    scaler = MinMaxScaler(feature_range=(0, 1))
    df["Scaled"] = scaler.fit_transform(df["Price"].values.reshape(-1, 1))


    If I configure my model as follows



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

    model = Sequential()
    model.add(LSTM(no_neuron, input_shape=(no_sample, timestep, no_feature)))
    model.add(Dense(1))
    model.compile(loss='mean_squared_error', optimizer='adam')


    Is that make sense to set my timestep=40, no_sample=140, no_feature=1 for my problem? and How to define batch size?



    Q3



    In my settings, I really need to predict the price from Apr 2018 until Dec 2018 is that possible to do even I have only data from Jan 2018 to Mar 2018?









    share







    New contributor




    Ratchainant Thammasudjarit is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
    Check out our Code of Conduct.







    $endgroup$















      0












      0








      0





      $begingroup$


      I follow several tutorials about LSTM multi-step forecast to solve my problem.



      My problem:
      I have a time series of price about 3 months (Jan 2018 to Mar 2018) and I assume there are seasonal. So, my time series behaviors are changed because of such seasonal effects.



      Q1
      Do I need to apply sliding window to prevent learning from too long sequence. For example, each training sample takes prices of previous 14 days to predict the price of next 14 days. In other words, my input is (t-1), (t-2), ... , (t-7) and my output is (t), (t+1), ... , (t+6).



      Q2
      I read this post and feel that it's interesting by just passing a whole sequence and let the model makes decision by itself using the advantage of setting return_sequence=True and stateful=True. However, I'm a newbie to LSTM so I could not have clear understand how to apply it to the model according to such post.



      Suppose my data look like this



      import pandas as pd
      import numpy as pd

      np.random.seed(40)
      df = pd.DataFrame({"Price":list(np.random.choice(range(1000, 1500), 14)) +
      list(np.random.choice(range(1500, 2000), 14)) +
      list(np.random.choice(range(1800, 2500), 14)) +
      list(np.random.choice(range(1200, 1500), 14)) +
      list(np.random.choice(range(1500, 2000), 14)) +
      list(np.random.choice(range(1800, 2300), 14)) +
      list(np.random.choice(range(2500, 3000), 14)) +
      list(np.random.choice(range(2400, 2800), 14)) +
      list(np.random.choice(range(2600, 3000), 14)) +
      list(np.random.choice(range(2300, 2700), 14))},
      index=pd.date_range(start="2018-03-25", periods=140))

      df["Price"] = df["Price"].astype(float)
      scaler = MinMaxScaler(feature_range=(0, 1))
      df["Scaled"] = scaler.fit_transform(df["Price"].values.reshape(-1, 1))


      If I configure my model as follows



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

      model = Sequential()
      model.add(LSTM(no_neuron, input_shape=(no_sample, timestep, no_feature)))
      model.add(Dense(1))
      model.compile(loss='mean_squared_error', optimizer='adam')


      Is that make sense to set my timestep=40, no_sample=140, no_feature=1 for my problem? and How to define batch size?



      Q3



      In my settings, I really need to predict the price from Apr 2018 until Dec 2018 is that possible to do even I have only data from Jan 2018 to Mar 2018?









      share







      New contributor




      Ratchainant Thammasudjarit is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
      Check out our Code of Conduct.







      $endgroup$




      I follow several tutorials about LSTM multi-step forecast to solve my problem.



      My problem:
      I have a time series of price about 3 months (Jan 2018 to Mar 2018) and I assume there are seasonal. So, my time series behaviors are changed because of such seasonal effects.



      Q1
      Do I need to apply sliding window to prevent learning from too long sequence. For example, each training sample takes prices of previous 14 days to predict the price of next 14 days. In other words, my input is (t-1), (t-2), ... , (t-7) and my output is (t), (t+1), ... , (t+6).



      Q2
      I read this post and feel that it's interesting by just passing a whole sequence and let the model makes decision by itself using the advantage of setting return_sequence=True and stateful=True. However, I'm a newbie to LSTM so I could not have clear understand how to apply it to the model according to such post.



      Suppose my data look like this



      import pandas as pd
      import numpy as pd

      np.random.seed(40)
      df = pd.DataFrame({"Price":list(np.random.choice(range(1000, 1500), 14)) +
      list(np.random.choice(range(1500, 2000), 14)) +
      list(np.random.choice(range(1800, 2500), 14)) +
      list(np.random.choice(range(1200, 1500), 14)) +
      list(np.random.choice(range(1500, 2000), 14)) +
      list(np.random.choice(range(1800, 2300), 14)) +
      list(np.random.choice(range(2500, 3000), 14)) +
      list(np.random.choice(range(2400, 2800), 14)) +
      list(np.random.choice(range(2600, 3000), 14)) +
      list(np.random.choice(range(2300, 2700), 14))},
      index=pd.date_range(start="2018-03-25", periods=140))

      df["Price"] = df["Price"].astype(float)
      scaler = MinMaxScaler(feature_range=(0, 1))
      df["Scaled"] = scaler.fit_transform(df["Price"].values.reshape(-1, 1))


      If I configure my model as follows



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

      model = Sequential()
      model.add(LSTM(no_neuron, input_shape=(no_sample, timestep, no_feature)))
      model.add(Dense(1))
      model.compile(loss='mean_squared_error', optimizer='adam')


      Is that make sense to set my timestep=40, no_sample=140, no_feature=1 for my problem? and How to define batch size?



      Q3



      In my settings, I really need to predict the price from Apr 2018 until Dec 2018 is that possible to do even I have only data from Jan 2018 to Mar 2018?







      machine-learning deep-learning keras time-series lstm





      share







      New contributor




      Ratchainant Thammasudjarit is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
      Check out our Code of Conduct.










      share







      New contributor




      Ratchainant Thammasudjarit is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
      Check out our Code of Conduct.








      share



      share






      New contributor




      Ratchainant Thammasudjarit is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
      Check out our Code of Conduct.









      asked 5 mins ago









      Ratchainant ThammasudjaritRatchainant Thammasudjarit

      101




      101




      New contributor




      Ratchainant Thammasudjarit is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
      Check out our Code of Conduct.





      New contributor





      Ratchainant Thammasudjarit is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
      Check out our Code of Conduct.






      Ratchainant Thammasudjarit is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
      Check out our Code of Conduct.






















          0






          active

          oldest

          votes











          Your Answer





          StackExchange.ifUsing("editor", function () {
          return StackExchange.using("mathjaxEditing", function () {
          StackExchange.MarkdownEditor.creationCallbacks.add(function (editor, postfix) {
          StackExchange.mathjaxEditing.prepareWmdForMathJax(editor, postfix, [["$", "$"], ["\\(","\\)"]]);
          });
          });
          }, "mathjax-editing");

          StackExchange.ready(function() {
          var channelOptions = {
          tags: "".split(" "),
          id: "557"
          };
          initTagRenderer("".split(" "), "".split(" "), channelOptions);

          StackExchange.using("externalEditor", function() {
          // Have to fire editor after snippets, if snippets enabled
          if (StackExchange.settings.snippets.snippetsEnabled) {
          StackExchange.using("snippets", function() {
          createEditor();
          });
          }
          else {
          createEditor();
          }
          });

          function createEditor() {
          StackExchange.prepareEditor({
          heartbeatType: 'answer',
          autoActivateHeartbeat: false,
          convertImagesToLinks: false,
          noModals: true,
          showLowRepImageUploadWarning: true,
          reputationToPostImages: null,
          bindNavPrevention: true,
          postfix: "",
          imageUploader: {
          brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
          contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
          allowUrls: true
          },
          onDemand: true,
          discardSelector: ".discard-answer"
          ,immediatelyShowMarkdownHelp:true
          });


          }
          });






          Ratchainant Thammasudjarit is a new contributor. Be nice, and check out our Code of Conduct.










          draft saved

          draft discarded


















          StackExchange.ready(
          function () {
          StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fdatascience.stackexchange.com%2fquestions%2f45558%2flstm-multi-state-forecast%23new-answer', 'question_page');
          }
          );

          Post as a guest















          Required, but never shown

























          0






          active

          oldest

          votes








          0






          active

          oldest

          votes









          active

          oldest

          votes






          active

          oldest

          votes








          Ratchainant Thammasudjarit is a new contributor. Be nice, and check out our Code of Conduct.










          draft saved

          draft discarded


















          Ratchainant Thammasudjarit is a new contributor. Be nice, and check out our Code of Conduct.













          Ratchainant Thammasudjarit is a new contributor. Be nice, and check out our Code of Conduct.












          Ratchainant Thammasudjarit is a new contributor. Be nice, and check out our Code of Conduct.
















          Thanks for contributing an answer to Data Science Stack Exchange!


          • Please be sure to answer the question. Provide details and share your research!

          But avoid



          • Asking for help, clarification, or responding to other answers.

          • Making statements based on opinion; back them up with references or personal experience.


          Use MathJax to format equations. MathJax reference.


          To learn more, see our tips on writing great answers.




          draft saved


          draft discarded














          StackExchange.ready(
          function () {
          StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fdatascience.stackexchange.com%2fquestions%2f45558%2flstm-multi-state-forecast%23new-answer', 'question_page');
          }
          );

          Post as a guest















          Required, but never shown





















































          Required, but never shown














          Required, but never shown












          Required, but never shown







          Required, but never shown

































          Required, but never shown














          Required, but never shown












          Required, but never shown







          Required, but never shown







          Popular posts from this blog

          How to label and detect the document text images

          Vallis Paradisi

          Tabula Rosettana