Cannot find the electricity dataset

[binh0206]

I am a newbie in R. I hope to receive the help. When I ran the electricity example code, I got an error for electricity dataset, in which electricity::electricity_hourly. I tried to apply install.packages("electricity"), but the error persists. I search in this repository, but I see no questions related. Therefore, I assume it is because of my set up. I use R version 4.2.3, R-Studio 2023.03.0 build 386. I had all required library installed. Would you mind helping to fix this error? Thank you so much!

[dfalbel]

Hi, sorry, you currently need to install it from github: remotes::install_github("dfalbel/electricity")

[binh0206]

Thank you so much for your rapid response! The error is fixed. However, I apologize to bother you one more time for step_include_roles(). The error is Error in step_include_roles(.) : could not find function "step_include_roles", I saw this function showed only one time in this entire repository without function definition. Would you please help? Thank you so much!

[dfalbel]

What happens if you just discard it? I think I might have experimented with it but ended up removing.

[binh0206]

Thanks for your response! Then, another error pops up is for function called tft. I replaced tft with tft::temporal_fusion_transformer. then, it causes another error such as

Error in fun(..., .envir = .envir): ! Could not evaluate cli {} expression: class(dataset). Caused by error in eval(expr, envir = envir): ! object 'dataset' not found

I think it is because the parameters cannot be rec and elec_train, but it should be spec only with spec <- tft_dataset_spec(rec, elec_train).

Is it correct?

[dfalbel]

Sorry, I would need to run the example to figure out. Have you tried looking at the differences between that electricity example and the one in: https://mlverse.github.io/tft/articles/Getting-started.html

Probably should help to know what's happening. Unfortunatelly this package is not in the state we wanted. There are some missing parts and the documentation is still sparse.

[binh0206]

Thanks for your response! Yes, I ran the code in Getting-started.html before, and it worked great without any errors. Therefore, I thought the reason is because the code has not converted elec_train to torch::dataset() using tft_dataset_spec.

Also, in tft::temporal_fusion_transformer, I think function tft::temporal_fusion_transformer only needs torch::dataset(), not both rec and elec_train.

I will try to work around to find the way to fix that error based on Gettiing-started.html. If you have an opportunity to test the code again and fix it, please let me know! Thank you so much!