Valeriy
05/31/2024, 1:51 PMFederica MEMBRETTI
05/31/2024, 3:05 PMfilter_nan_and_inf
arg to False."
Here is my code:
-----------------------------
forecast_horizon = 52
freq = "W-SUN"
config_ray = {
"stat_exog_list": stat_exog_list,
"hist_exog_list": hist_exog_list,
"input_size": 2*forecast_horizon,
"max_steps": 1500,
"val_check_steps": 1500,
"hidden_size": tune.choice([256, 300]),
"windows_batch_size": 1024,
"inference_windows_batch_size": 1024,
"n_head": 2,
"learning_rate": tune.loguniform(1e-3, 0.1),
"scaler_type": tune.choice([ "standard", "minmax1", "robust"]),
"dropout": tune.uniform(0.0, 0.2),
"start_padding_enabled": True,
}
models_ray = [
AutoTFT(
h=forecast_horizon,
loss= DistributionLoss("StudentT", num_samples=1000, level=[20, 40, 60, 80, 100]),
config=config_ray,
search_alg=HyperOptSearch(),
backend="ray",
num_samples=5,
verbose=3,
gpus=1
)
]
nf = NeuralForecast(models=models_ray, freq=freq)
nf.fit(df=df, static_df=static_df, val_size=forecast_horizon)
-----------------------------------------------------------------------------------
Do you know what may have happened here?
Any help would be really appreciated! Thank you :)DJ Passey
05/31/2024, 7:25 PMhistoric_endog
argument to a neural forecast dataframe and pass it to the right NeuralForecast
method with in the NeuralForecastAdapter._predict
function in the linked code above.Ruben
06/01/2024, 8:18 AMValeriy
06/03/2024, 2:00 PMValeriy
06/04/2024, 9:50 AMChristian Silva
06/05/2024, 9:28 PMAman Singh
06/07/2024, 9:58 AMAman Singh
06/08/2024, 6:17 AMArun Bharadwaj
06/12/2024, 10:08 AM---> 10 from neuralforecast.models import SOFTS, PatchTST, TSMixer, iTransformer
11
ImportError: cannot import name 'SOFTS' from 'neuralforecast.models' (/usr/local/lib/python3.10/dist-packages/neuralforecast/models/__init__.py)
Martin Drozda
06/12/2024, 2:06 PMpredictions = nf.predict(df=df_val, sort_df=False, verbose=True)
How to get predictions for the complete df_val, not just one single prediction? Why is the single prediction not moved by encoder length, i.e. where does predict() get past time series values, if it returns prediction for the first item in df_val? Why does the nixtla implementation of TFT have less input parameters than pytorch_forecasting implementation? What is the difference between these two implementations?Naren Castellon
06/12/2024, 3:42 PMDistributionLoss()
parameters and the forecast gives me two results, for example LSTM and LSTM-median, which one should I use, or can I consider it as another model?
I can consider it as another model, because when I use cross validation, to evaluate the model, it considers the LSTM and the LSTM-median as another model!!!
Another question, does neuralforecast have any parameters similar to mlforecast's target_transforms()
that can transform the variable?Zncheon94
06/12/2024, 4:01 PMMl Club
06/13/2024, 6:57 AMThiago Theiry de Oliveira
06/17/2024, 10:16 PMTina Sedaghat
06/18/2024, 4:39 PMVidar Ingason
06/18/2024, 5:30 PMHua Tang
06/19/2024, 6:13 PMHua Tang
06/20/2024, 6:02 AMHuseyn Zeynalov
06/20/2024, 7:39 AMGiovanni Perri
06/21/2024, 9:43 AMBill
06/22/2024, 3:40 AMBrijesh Mantri
06/24/2024, 3:11 PMTal Zaquin
06/25/2024, 9:17 AMavailable_mask
column. But I couldn't find any documentation about it.
a. Should I add it to my train df?
b. Is there a need to define it somewhere in the model or the fit command?
c. I also have some exogenous features that I would like to use, but they also have missing values in certain timestamps (the same as the target). Should I leave it as nan (let's say I'm using the TSMixerx model)?
2. This is more of a theoretical question, and continuing with the use of the TSMixerx model: I have multiple time series that I want to run multivariate forecasting. However, due to complexities in data acquisition, the time series have different lengths, and they do not start or end on the same timestamp. The only good thing about it is that most of the time stamps are found in all of the series.
a. Could you suggest a way to deal with it?
b. Can I do multivariate forecasting, or should I go for univariate forecasting and choose a single model for each series?Vidar Ingason
06/25/2024, 12:49 PMmodel = AutoNHITS(h=12,
loss=MAE(),
config=nhits_config,
search_alg=HyperOptSearch(),
backend='ray',
num_samples=10)
nf = NeuralForecast(models=[model], freq='M')
nf.fit(df=Y_df, val_size=24)
Banafsheh Nikbakht
06/25/2024, 4:38 PMYag ger Phone
06/26/2024, 8:08 AMJasmine Rienecker
06/27/2024, 11:42 PMJonathan
06/28/2024, 12:31 PMD N
07/03/2024, 3:45 PM