search

andrewzm / deepIDE

Deep IDE Models for Spatio-Temporal Forecasting
21 stars 9 forks source link

How to predict data with independent variables (influencing factors) #4

Open 19211020075 opened 2 years ago

19211020075 commented 2 years ago

Hello, If I want to predict data with independent variables (influencing factors), how do I get the coefficients of the independent variables and and get the statistical significance? Looking forward to your reply,Best wishes! Lino

andrewzm commented 2 years ago

Hi Lino, I'm not sure how this relates to deepIDE. Does your question relate to the IDE package?

On Mon, May 16, 2022 at 4:30 PM 19211020075 @.***> wrote:

Hello, If I want to predict data with independent variables (influencing factors), how do I get the coefficients of the independent variables and and get the statistical significance? Looking forward to your reply,Best wishes! Lino

— Reply to this email directly, view it on GitHub https://github.com/andrewzm/deepIDE/issues/4, or unsubscribe https://github.com/notifications/unsubscribe-auth/AABEU3PNJ3ME3ERBP7X6HZDVKHTPVANCNFSM5WAM3OOQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>

19211020075 commented 2 years ago

I would like to use the deepIDE to conduct the models. Your data had only dependent variables, no independent variables included (i.e. SST and radar). However, My data have independent variables (influencing factors), and I want to predict the dependent variables and explore the relationship between the influencing factors and the dependent variables. Whether the relevant coefficients can be obtained by the deepIDE. If can, where and how should I tweak the code in the model building?
Thank you!

andrewzm commented 2 years ago

There are no covariates in the deepIDE model, the way I designed it... it's just a model for a (zero-mean) spatio-temporal process. If this is evolving on top of some trend (which I take to imply you are modelling as X.beta since you mention coefficients?) then you need to alter the inference mechanism. Specifically, you need to develop an inferential procedure for the model

Z_t,i = Ytilde_t(s_i) + epsilon_t,i Ytilde_t(s_i) = X_t(s_i).beta + Y_t(s_i) Y_t(.) ~ deepIDE()

and estimate those beta parameters. I don't think it should be too hard, but it would entail a bit of work.

On Tue, May 17, 2022 at 11:36 AM 19211020075 @.***> wrote:

I would like to use the deepIDE to conduct the models. Your data had only dependent variables, no independent variables included (i.e. SST and radar). However, My data have independent variables (influencing factors), and I want to predict the dependent variables and explore the relationship between the influencing factors and the dependent variables. Whether the relevant coefficients can be obtained by the deepIDE. If can, where and how should I tweak the code in the model building? Thank you!

— Reply to this email directly, view it on GitHub https://github.com/andrewzm/deepIDE/issues/4#issuecomment-1128308249, or unsubscribe https://github.com/notifications/unsubscribe-auth/AABEU3LJWUXTYKPDIGHSVOTVKLZ2DANCNFSM5WAM3OOQ . You are receiving this because you commented.Message ID: @.***>