6 Pull requests merged by 4 people

• Fixes bug when using historical forecast with discrete lags < -1

  #2715 merged Mar 7, 2025

• feat: adding support for getting estimator based on quantile

  #2716 merged Mar 7, 2025

• Fix/inverse transform for static covariate with single category across series

  #2710 merged Mar 7, 2025

• Catboost native multi-output with RMSE

  #2659 merged Mar 7, 2025

• Fix wrong call to TimeSeries.from_series()

  #2719 merged Mar 6, 2025

• Bump jinja2 from 3.1.5 to 3.1.6 in /requirements

  #2717 merged Mar 6, 2025

2 Pull requests opened by 1 person

• feat: allow stride=forecast_horizon for anomaly forecasting model

  #2709 opened Mar 3, 2025

• Fix/code maintenance after dependencies update

  #2722 opened Mar 7, 2025

16 Issues closed by 4 people

• [QUESTION] Single lag values in linear regression causing trend and seasonal backtests to have different periods when retrain=False

  #2697 closed Mar 7, 2025

• [BUG] Can't access underlying models in multi-quantile regression

  #2596 closed Mar 7, 2025

• [BUG] A StaticCovariatesTransformer fit to multiple categorical static covariates with the same constant value is uninvertable

  #2705 closed Mar 7, 2025

• Make CatboostModel use native Catboost multi-output when possible

  #1306 closed Mar 7, 2025

• [QUESTION] Can we use a PyTorch Model (e.g `Transformer`) as a `RegressionModel`?

  #2720 closed Mar 7, 2025

• `TimeSeries.from_series()` do not work when called with a `pd.Index`

  #2718 closed Mar 6, 2025

• [QUESTION] Train and inference TFTModel with past_covariates values only

  #2629 closed Mar 5, 2025

• [BUG] - TCNModel - save/load does not work anymore - darts [0.32.0]

  #2630 closed Mar 5, 2025

• Cannot import darts.timeseries.slice_intersect(series)

  #2714 closed Mar 5, 2025

• TCNModel loading data when training

  #2712 closed Mar 5, 2025

• [BUG] transformer Scaller does not work with `add_encoders`

  #2713 closed Mar 4, 2025

• Is It Possible to Deploy Darts Model in a TinyML Setup?

  #2706 closed Mar 4, 2025

• [QUESTION] Model size and trainable parameters

  #2711 closed Mar 4, 2025

• Calling historical_forecast() after loading a model

  #2708 closed Mar 4, 2025

• Error in Historical Forecast Using Static Covariates Trained by fit_from_dataset

  #2110 closed Mar 3, 2025

• TiDE Model Stops At A Specific Epoch

  #2496 closed Mar 3, 2025

2 Issues opened by 2 people

• [QUESTION] Is there any way to make backtesting with multi-shift/dynamic future covariates?

  #2721 opened Mar 7, 2025

• How to model with only 6 months of data per year?

  #2707 opened Mar 3, 2025

16 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• `to_dataframe(backend)` and `to_series(backend)`

  #2701 commented on Mar 7, 2025 • 26 new comments

• Feat/test optional dep (onnx, ray, optuna)

  #2702 commented on Mar 7, 2025 • 11 new comments

• Add support for TFT feature importance over time

  #2685 commented on Mar 3, 2025 • 0 new comments

• Add MFLES from StatsForecast

  #2636 commented on Mar 3, 2025 • 0 new comments

• TimeSeries.max(axis=0), .min(axis=0) return a meaningless index

  #2696 commented on Mar 3, 2025 • 0 new comments

• [BUG]: NaiveEnsemble().historical_forecast() cannot handle multiple past_covariates

  #2664 commented on Mar 4, 2025 • 0 new comments

• [BUG] Several issues with NLinear with `normalize=True`

  #2699 commented on Mar 4, 2025 • 0 new comments

• [QUESTION] Error during inference with NBEATS model.

  #2642 commented on Mar 4, 2025 • 0 new comments

• [QUESTION] What's the correct way to train XGBoost on a multi-source dataset?

  #2570 commented on Mar 5, 2025 • 0 new comments

• [QUESTION] How to evaluate training loss before the first epoch?

  #2704 commented on Mar 5, 2025 • 0 new comments

• [FEAT] Add tsmixer-basic

  #2510 commented on Mar 5, 2025 • 0 new comments

• Help Reproducing the TIDE Paper

  #2644 commented on Mar 5, 2025 • 0 new comments

• Efficient and Reliable Ways to Diagnose Overestimation in LightGBM Model for Load Forecasting

  #2694 commented on Mar 7, 2025 • 0 new comments

• Imp/tsmixer basic

  #2555 commented on Mar 5, 2025 • 0 new comments

• Feat/strided dataset for torch and regression models

  #2624 commented on Mar 3, 2025 • 0 new comments

• Feature/metadata

  #2656 commented on Mar 6, 2025 • 0 new comments