Mikhail Sarafanov

Some bugs arised during time series benchmark experiments (the post will be updated periodically): ``` OptimisationTimer - Composition time: 1.8 min Exception "no default __reduce__ due to non-trivial __cinit__" raised....

A lot of messages appear during launch Example: ``` from fedot.api.main import Fedot model = Fedot(problem='regression', timeout=1) # fit model with AutoML train_columns = df.columns[:-1] obtained_pipeline = model.fit(features=df[train_columns], target=df['target']) ```...

The framework has the `multi_ts` data type. However, when developing functionality for multimodal data processing, the `data_source_ts` node is put on a par with the `muti_ts` data. There is a...

Error stacktrace:  ``` ValueError: The first guess on the deviance function returned a nan. This could be a boundary problem and should be reported. ``` Example to run: https://github.com/nccr-itmo/FEDOT/blob/master/cases/multivariate_ts_forecasting.py...

The split factor for the time series is now ignored. There is a need to implement time series in-sample forecasting to use this info during validation ``` _sample_split_ratio_for_tasks = {...

Need to add tests for cross validation on time series and tables. Metrics should be measured, and monitor how models will predict on synthetic data. Example: generate synthetic dataset with...

The alternative title for the future PR on this issue is "Batch-support implementation for image processing" FEDOT currently has a separate DataTypesEnum.image data type. It loads all matrices into an...

Now the time series prediction model predicts for a given number of elements (forecast horizon). But for some models, training for long horizons may be ineffective. So it is possible...

Mutation of hyperparameters with correction

1

It is required to implement intelligent mutation of hyperparameters in operations. Therefore, we need to know the intervals of hyperparameter values (min and max). This cannot be done in the...

Forecasting smoothed time series

3

A possible improvement of the time series smoothing operation is proposed. Below is example of applying smoothed operation  Currently, the implementation of smoothing allows modifying only the "features" time...