neuralforecast
neuralforecast copied to clipboard
Nixtla
Scalable and user friendly neural :brain: forecasting algorithms.
### Description I guess an actual TFT can handle categorical features with embeddings, does nixtla have something similar or it always needs to be done manually with label encodings? ###...
### Description Currently, there isn’t a way to specify a ray scheduler, neuralforecast just uses the default FIFO one. If you want to use a sampling algorithm like BOHB, you...
Solves predict_insample for issue #1346
### What happened + What you expected to happen ## Description The current implementation of `isqf_domain_map` function applies `F.softplus()` to all quantile knots, including the first one. This is incorrect...
Dear sir, Here is my code, which uses LSTM for multi-step forecasting. I have set the stat_exog_list parameter. However, when I use the predict_insample method, I cannot find a way...
### Description ### Feature Request: Add `df_val` to `.fit()` for validation on disjoint `unique_id`s during hyperparameter tuning In my use case, each `unique_id` represents a short time series — a...
### What happened + What you expected to happen I suspect a bug, or at least an unintentional change introduced by the 3.0.0 release, which impacts forecasts when using a...
Currently, we create the maximum number of training windows, meaning that we might have windows with only 1 available insample data point and 1 available outsample data point. These are...
### What happened + What you expected to happen When using Nixtla’s RNN-Direct or LSTM-Direct classes to forecast four steps at once, the resulting mean squared error is consistently identical...
### What happened + What you expected to happen CV with refit=True on Pandas input DataFrame fails because there seems to be a mismatch in associated dates when the prediciton...
