The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.
nixtlar
provides an R interface to Nixtla’s TimeGPT, a generative
pre-trained forecasting model for time series data. TimeGPT
is the first foundation model capable of producing accurate forecasts
for new time series not seen during training, using only its historical
values as inputs. TimeGPT
can also be used for other time
series related tasks, such as anomaly detection and cross-validation.
Here we explain how to get started with TimeGPT
in R and
give a quick overview of the main features of nixtlar
.
First, you need to set up your API key. An API key is a string of
characters that allows you to authenticate your requests when using
TimeGPT
via nixtlar
. This API key needs to be
provided by Nixtla, so if you don’t have one, please request one here.
When using nixtlar
, there are two ways of setting up
your API key:
nixtla_client_setup
functionnixtlar
has a function to easily set up your API key for
your current R session. Simply call
Keep in mind that if you close your R session or you re-start it, then you’ll need to set up your API key again.
When using Azure, you also need to add the base_ur
parameter to the nixtla_client_setup
function.
For a more persistent method that can be used across different
projects, set up your API key as environment variable. To do this, first
load the usethis
package.
This will open your .Reviron
file. Place your API key
here and named it NIXTLA_API_KEY
.
You’ll need to restart R for changes to take effect. Keep in mind
that modifying the .Renviron
file affects all of your R
sessions, so if you’re not comfortable with this, use the
nixtla_client_setup
function instead.
If you are using Azure, you also need to specify the
NIXTLA_BASE_URL
.
For details on how to set up your API key, check out the Setting Up Your API Key vignette. To learn more about how to use Azure, please refer to the TimeGEN-1 Quickstart (Azure).
If you want to validate your API key, call
nixtla_validate_api_key
.
You don’t need to validate your API key every time you set it up,
only when you want to check if it’s valid. The
nixtla_validate_api_key
will return TRUE
if
you API key is valid, and FALSE
otherwise.
Once your API key has been set up, you’re ready to use
TimeGPT
. Here we’ll show you how this is done using a
dataset that contains prices of different electricity markets.
df <- nixtlar::electricity
head(df)
#> unique_id ds y
#> 1 BE 2016-10-22 00:00:00 70.00
#> 2 BE 2016-10-22 01:00:00 37.10
#> 3 BE 2016-10-22 02:00:00 37.10
#> 4 BE 2016-10-22 03:00:00 44.75
#> 5 BE 2016-10-22 04:00:00 37.10
#> 6 BE 2016-10-22 05:00:00 35.61
To generate a forecast for this dataset, use
nixtla_client_forecast
. Default names for the time and the
target columns are ds
and y
. If your time and
target columns have different names, specify them with
time_col
and target_col
. Since it has multiple
ids (one for every electricity market), you’ll need to specify the name
of the column that contains the ids, which in this case is
unique_id
. To do this, simply use
id_col="unique_id"
. You can also choose confidence levels
(0-100) for prediction intervals with level
.
nixtla_client_fcst <- nixtla_client_forecast(df, h = 8, level = c(80,95))
#> Frequency chosen: h
head(nixtla_client_fcst)
#> unique_id ds TimeGPT TimeGPT-lo-95 TimeGPT-lo-80
#> 1 BE 2016-12-31 00:00:00 45.19045 30.49691 35.50842
#> 2 BE 2016-12-31 01:00:00 43.24445 28.96423 35.37463
#> 3 BE 2016-12-31 02:00:00 41.95839 27.06667 35.34079
#> 4 BE 2016-12-31 03:00:00 39.79649 27.96751 32.32625
#> 5 BE 2016-12-31 04:00:00 39.20454 24.66072 30.99895
#> 6 BE 2016-12-31 05:00:00 40.10878 23.05056 32.43504
#> TimeGPT-hi-80 TimeGPT-hi-95
#> 1 54.87248 59.88399
#> 2 51.11427 57.52467
#> 3 48.57599 56.85011
#> 4 47.26672 51.62546
#> 5 47.41012 53.74836
#> 6 47.78252 57.16700
nixtlar
includes a function to plot the historical data
and any output from nixtla_client_forecast
,
nixtla_client_historic
,
nixtla_client_anomaly_detection
and
nixtla_client_cross_validation
. If you have long series,
you can use max_insample_length
to only plot the last N
historical values (the forecast will always be plotted in full).
These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.