Does anyone have experience with using the notebook from Advanced Feature Selection with R | DataRobot Community?
Two issues I am having with it:
1. The ListModels function is throwing an error, the traceback is below.
2. I am expecting to run into problems modifying it to work with Time Series projects. So if anybody has advice with feature selection for TS models I would appreciate it.
Error in curl::curl_fetch_memory(url, handle = handle) : Protocol "" not supported or disabled in libcurl
9.
curl::curl_fetch_memory(url, handle = handle)
8.
request_fetch.write_memory(req$output, req$url, handle)
7.
request_fetch(req$output, req$url, handle)
6.
request_perform(req, hu$handle$handle)
5.
(function (url = NULL, config = list(), ..., handle = NULL) { hu <- handle_url(handle, url, ...) req <- request_build("GET", hu$url, as.request(config), ...) ...
4.
do.call(requestMethod, args)
3.
MakeDataRobotRequest(httr::GET, routeString, addUrl = addUrl, returnRawResponse = returnRawResponse, ...)
2.
DataRobotGET(routeString, simplify = FALSE, query = params)
1.
ListModels(project)
My bad, I read your or as an and.
I have tried both types of projects, none worked, all with the same error/traceback. Thanks for taking a look at this I appreciate it!
You're welcome! Thanks for your patience as we continue to work on this.
Can you confirm if the functionality at https://cran.r-project.org/web/packages/datarobot/vignettes/TimeSeries.html is working for you? This isn't about `ListModels()` exactly but I want to diagnose if there are multiple issues for you and your Time Series projects or if we should just isolate on `ListModels()`.
So I ran from the vignette:
data <- read.csv(system.file("extdata", "multiseries.csv", package = "datarobot"))
partition <- CreateDatetimePartitionSpecification(datetimePartitionColumn = "timestamp",
useTimeSeries = TRUE,
multiseriesIdColumns = "series_id")
project <- StartProject(data,
projectName = "test-TimeSeries",
target = "target",
partition = partition,
metric = "RMSE",
mode = AutopilotMode$Manual,
targetType = "Regression")
which threw
Project test-TimeSeries creation requested, awaiting creation
No encoding supplied: defaulting to UTF-8.
Error in if (grepl("//$", url) || grepl("\\s", url) || !grepl("/", url)) { :
missing value where TRUE/FALSE needed
The data file had some strange corruption.
But I adjusted the script to work with the data used in the TS project I am working on (via local import) and it gave the same error.
Hi @GR - are you still struggling with this?
I managed to get the python script working (from the original article). Still cannot get the R function 'ListModels" to work.