Closed jmrichardson closed 4 years ago
I'm sorry, I don't understand this question. How are you using mcfly?
@cantino I am currently using mcfly the way that is documented in the examples and everything is working great. In my use case, I am feeding mcfly with financial price time series data for example MSFT open, high, low, close on a minute basis. However, I also have contextual data for example daily or weekly data. For example, I may have an interest rate value that I would like to feed into the model that would change on an irregular basis. The seglearn package refers to this as contextual data and has the ability to include in the model (here). Instead of having a feature called intrate that would be constant on a minute basis, is there a way to feed this scalar into the model using mcfly? After some research online, it appears that some folks engineer it into the layers:
Thanks
I'm pretty sure you're talking about a different mcfly project.
That makes more sense ;)
On Sat, Jul 18, 2020 at 12:31 PM jmrichardson notifications@github.com wrote:
whooops... sorry, yes wrong mcfly:
https://github.com/NLeSC/mcfly
Thanks anyways :)
— You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub https://github.com/cantino/mcfly/issues/92#issuecomment-660530896, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAUO67DYST4TXNAIZMRFD3R4H2CFANCNFSM4O7FYMGA .
Thank you for mcfly, it is very helpful. I have a question regarding adding scalar inputs. I would like to to include contextual data in each subsequence without just making a feature of all the same value. Is this possible? For example, lets say I have a training set of shape:
20, 50, 5
For each of the 20 samples, I would like to include a scalar value indicating context for example (0 or 1).
Thanks for any advice