Examples

We have prepared a set of tutorials for an easy introduction:

01. Get started

• Creating TSDataset and time series plotting
• Forecast single time series - Simple forecast, Prophet, Catboost
• Forecast multiple time series
• Pipeline

02. Backtest

• What is backtest and how it works
• How to run a validation
• Validation visualisation

03. EDA

• Visualization
  • Plot
  • Partial autocorrelation
  • Cross-correlation
  • Correlation heatmap
  • Distribution
• Outliers
  • Median method
  • Density method
• Change Points
  • Change points plot
  • Interactive change points plot

04. Regressors and exogenous data

• What is regressor?
  • What is exogenous data?
• Dataset
  • Loading Dataset
  • EDA
• Forecast with regressors

05. Custom model and transform

• What is Transform and how it works
• Custom Transform
  • Per-segment Custom Transform
  • Multi-segment Custom Transform
• Custom Model

06. Deep learning models

• Creating TSDataset
• Architecture
• Testing models
  • DeepAR
  • TFT
  • Simple Model

07. Ensembles

• VotingEnsemble
• StackingEnsemble

08. Outliers

• Point outliers
  • Median method
  • Density method
  • Prediction interval method
  • Histogram method
• Sequence outliers
• Interactive visualization
• Outliers imputation

09. Forecasting strategies

• Imports and constants
• Load dataset
• Recursive strategy
  • AutoRegressivePipeline
• Direct strategy
  • Pipeline
  • DirectEnsemble
  • assemble_pipelines + DirectEnsemble
• Summary

10. Forecast interpretation

• Forecast decomposition
  • CatBoost
  • SARIMAX
  • BATS
  • In-sample and out-of-sample decomposition
• Accessing target components
• Regressors relevance
  • Feature relevance
  • Components relevance

11. Clustering

• Clustering pipeline
• Custom Distance
• Visualisation

12. AutoML script

• Auto pipeline search

13. AutoML notebook

• Hyperparameters tuning
  • How Tune works
  • Example
• General AutoML
  • How Auto works
  • Example

14. Hyperparameter search

• Optuna
• WandB sweeps example based on Hydra

15. Inference: using saved pipeline on a new data

• Fitting and saving pipeline
• Using saved pipeline on a new data

16. Hierarchical time series

• Hierarchical time series
• Hierarchical structure
• Reconciliation methods
• Exogenous variables for hierarchical forecasts

17. Classification

• Classification
  • Load Dataset
  • Feature extraction
  • Cross validation
• Predictability analysis
  • Load Dataset
  • Load pretrained analyzer
  • Analyze segments predictability

18. Feature selection

• Loading Dataset
• Feature selection methods
  • Intro to feature selection
  • TreeFeatureSelectionTransform
  • GaleShapleyFeatureSelectionTransform
  • MRMRFeatureSelectionTransform