Glossary

There are many disciplines involved in the world of time series forecasting and many use different names for the same/similar things. To make it easier for further package navigation, these are important terms mentioned throughout the package.

Term

Meaning

Synonyms

Wrapper

Convenient adapter to third party estimators following Sklearn API enabling time series forecasting

Horizon

Number of datapoints within defined data frequency that are predicted

Steps-ahead

Cross-validation

Procedure to run model’s fitting and predicting on several test/train data splits in order to assess model ability to predict out-of-sample

CV

Split

One of the train/test data subsamples (n_splits) created during Cross-validation (based on splitting strategy)

Target

Data we are trying to predict in form array_like, (1d)

Target vector, Target variable, Dependent variable, Label, Response variable, Explained variable, Outcome variable, Output variable, Endogenous variable, Y

Feature

Measurable property (e.g. is_holiday) of the target (e.g. sales) in form of pandas.Dataframe

Exogenous variable, Covariate, Regressor, Independent variable, Explanatory variable, Predictor variable, Input, X