Can we use a pandas dataframe with multiple dependent variables?

[thepraveen19]

I want to analyse the stock market data which has multiple parameters as input variables and one target variable. Is it possible to use NBEATS ??

[philipperemy]

https://github.com/philipperemy/n-beats/issues/60

@thepraveen19 the original paper of NBeats only considered 1-D time series and the target was X[t+1] when the context was X[t], X[t-1]..., X[t-N].

Two points:

• But in practice, you don't need to have both X in the predictors and the target.
• Also NBeats is a RNN but you technically don't need to use it as an RNN as well (you can forget the time dimension)

So it's definitely possible to use a Pandas DataFrame with multiple depend variables.

A simple way to do that:

https://github.com/philipperemy/n-beats/blob/master/examples/pandas_example.py

[thepraveen19]

Hi,

Thank you for the revert. A quick question, you mentioned that we can forget the time dimension. I don't understand how it works from the statistical standpoint. Is there a document or something that can help me understand how it functions?

Your help is much appreciated.

Thank you

Regards, Praveen

On Fri, 14 Jan 2022, 08:23 Philippe Rémy, @.***> wrote:

60 https://github.com/philipperemy/n-beats/issues/60

@thepraveen19 https://github.com/thepraveen19 the original paper of NBeats only considered 1-D time series and the target was X[t+1] when the context was X[t], X[t-1]..., X[t-N].

Two points:

• But in practice, you don't need to have both X in the predictors and the target.
• Also NBeats is a RNN but you technically don't need to use it as an RNN as well (you can forget the time dimension)

So it's definitely possible to use a Pandas DataFrame with multiple depend variables.

A simple way to do that:

https://github.com/philipperemy/n-beats/blob/master/examples/pandas_example.py

— Reply to this email directly, view it on GitHub https://github.com/philipperemy/n-beats/issues/60#issuecomment-1012906978, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKESZO4TMAZG2HQEQ57KMCDUV7MORANCNFSM5LADMNOQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you were mentioned.Message ID: @.***>

[philipperemy]

Hi,

I assumed that all the rows in the dataset are independent are permutable (meaning the order of your dataFrame does not matter). But if your dataset is indexed by time, yes we can take the time dimension into account. Let me work out another example for you.

[philipperemy]

Okay I updated the code here: https://github.com/philipperemy/n-beats/blob/master/examples/pandas_example.py

Including time.

[thepraveen19]

That is incredibly helpful...

I'm very grateful...

Thank you so much again ...

Regards, Praveen

On Fri, 14 Jan 2022, 08:58 Philippe Rémy, @.***> wrote:

Okay I updated the code here: https://github.com/philipperemy/n-beats/blob/master/examples/pandas_example.py

Including time.

— Reply to this email directly, view it on GitHub https://github.com/philipperemy/n-beats/issues/60#issuecomment-1012928940, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKESZO5JKM2CY7L7YKO7YDDUV7QR3ANCNFSM5LADMNOQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you were mentioned.Message ID: @.***>

[philipperemy]

Welcome!

[thepraveen19]

Hi,

It gives an error:

---

TypeError Traceback (most recent call last)

in () 5 output_dim=1, 6 nb_blocks_per_stack=1, ----> 7 backcast_length=timesteps) 8 TypeError: __init__() got an unexpected keyword argument 'output_dim' I checked the documentation, 'output_dim' is not built in. How can I Solve this?

[philipperemy]

@thepraveen19 right I need to push the latest version. Hang on.

[philipperemy]

Run this and try again.

pip install nbeats-keras --upgrade

[thepraveen19]

Awesome! You are amaziiiiiing..... I went through your website( sky sense)... you are doing great work.... Cheers!

[philipperemy]

Awesome! Thanks for the feedback :)