Data & Features¶
Transformers, feature engineering, stationarity, and pipeline composition. These notebooks show how to clean and resample time series, build lag and rolling features, apply differencing and decomposition, and wire transformers together with FeaturePipeline.
-
How to Handle Missing Data
Compare SimpleTimeImputer, SeasonalImputer, SimpleImputer, and TransformedSpaceKNNImputer on synthetic block and scattered gaps in monthly tourism data.
-
How to Use ColumnTransformer
Route columns through distinct transformers with ColumnTransformer, including remainder handling and automatic panel-aware column detection.
-
How to Clean Time Series Data
End-to-end data cleaning pipeline combining SimpleTimeImputer and SeasonalImputer for missing values with OutlierThresholdHandler for anomaly clipping.
-
Decomposition
Chain PolynomialTrendForecaster, PatternSeasonalityForecaster, and FourierSeasonalityForecaster inside DecompositionPipeline with component visualisation.
-
How to Compose Features with FeatureUnion
Combine lag features, rolling statistics, EMA, and scaling in parallel with FeatureUnion and automatic observation horizon resolution.
-
How to Tune Fourier Seasonality Terms
Explore how Fourier harmonic count affects seasonal fit quality, compare Fourier vs Pattern seasonality, and tune harmonics jointly with GridSearchCV.
-
How to Wrap Functions as Transformers
Wrap arbitrary polars or numpy operations as sklearn transformers with FunctionTransformer, supporting stateful warmup, inverse transforms, and pipelines.
-
How to Handle Long Series
Limit history with observation_horizon, weight recent errors with exponential decay, and downsample high-frequency data.
-
How to Handle Outliers in a Forecasting Pipeline
Detect and clip outliers with OutlierThresholdHandler and OutlierPercentileHandler, then see how outliers affect conformal prediction intervals.
-
How to Handle Short Series
Use Fourier seasonality, simple train/test splits, and panel pooling when individual series are too short for standard approaches.
-
How to Build a Feature Pipeline
Nest FeaturePipeline, FeatureUnion, and DecompositionPipeline for multi-level feature engineering with trend-season-residual decomposition.
-
How to Resample Time Series
Demonstrate Downsampler and Upsampler for changing time series frequency, including multivariate support, boundary settings, and round-trip information loss.
-
How to Apply Signal Processing Filters
Apply NumericalFilter (Butterworth, Chebyshev, Bessel), NumericalDifferentiator, and NumericalIntegrator for signal smoothing and rate-of-change extraction.
-
How to Use Scikit-learn Scalers
Wrap sklearn scalers (StandardScaler, MinMaxScaler, RobustScaler, PowerTransformer, PolynomialFeatures) for polars DataFrames with inverse transforms.
-
How to Apply Stationarity Transforms
Catalogue of variance-stabilising and detrending transforms: LogTransformer, BoxCox, SeasonalDifferencing, SeasonalReturn, and ASinh with inverse verification.
-
How to Align Exogenous Features Across Pipeline Steps
Control which step-indexed columns each direct-strategy estimator sees using the step_feature_alignment parameter of PointReductionForecaster.
-
How to Add Calendar, Fourier, and Holiday Features
Enrich your feature matrix with time-derived signals using CalendarFeatureTransformer, FourierFeatureTransformer, and HolidayFeatureTransformer.
-
How to Apply Window Transformations
Feature engineering with LagTransformer, RollingStatisticsTransformer, SlidingWindowFunctionTransformer, and ExponentialMovingAverage on time series data.