statworx / blog Goto Github PK

Hi @manueltilgner

First, thank you for your blogs!

I wanted to leave a comment about the time series forecasting blog at https://www.r-bloggers.com/2019/09/time-series-forecasting-with-random-forest/#google_vignette

X_test <- tax_ts_mbd[nrow(tax_ts_mbd), c(1:lag_order)]

I was wondering whether this should be instead

X_test <- tax_ts_mbd[nrow(tax_ts_mbd), -1]

because otherwise we keep the first column in the X_test, which is the target and should be out. We should only use the lagged versions for predicting. Or maybe I am just wrong :)

Anyway, happy if you could reply, otherwise, again, thanks for the neat tutorial!

Hi,

I found this project from a Statworx blog post:
https://www.statworx.com/en/blog/how-to-build-a-dashboard-in-python-plotly-dash-step-by-step-tutorial/#contents

It seems like the code referred to in the blog post is different to that in this repository. Perhaps an older version of the code is hosted here. Is there a python script available matching that shown in the blog post?

Hi, I love your https://github.com/STATWORX/blog/tree/master/flowchart it's really wonderful and what I've been looking for... however, I find the code a bit unexplained... I was wondering if you could perhaps add a bit more documentation, or perhaps just a small readme describing how to install or what functions to run.

Of course I think it's also powerful enough to convert into a package, but that's neither here nor there :)

Thanks!

Unfortunately link to GH repos in this blog post does not work anymore so I am filling issue here.

I briefly went trough the code of subset benchmark only and spotted potential issue in filtering queries. The way you are querying data.table makes it to fall back to data.frame way of doing subset. I am not sure if it makes big difference here, but it is definitely not the way to do that in data.table.
I explained this difference in "benchmarking best practices" document just now, you can preview it
Rdatatable/data.table@2556b05