# CartoBoost Docs This is a machine-readable index of the user-facing documentation. Developers and data scientists should begin with the docs home and getting-started guide; the website navigation remains the canonical human reading order. ## Main Entry Points - [Docs home](./index.md) - Versioned docs are built from `../docs-versions.json`; `latest` maps to the current docs, and retained release docs are added only for real tagged versions listed in that manifest. - [Getting started](./getting-started.md) - [Python estimator numeric input and missing-value contract](./user-guide/python-estimator.md) - [Choose a model](./user-guide/model-types.md) - [Stable estimators and supported model registry](./user-guide/model-types.md) - [Spatial econometrics models](./user-guide/spatial-econometrics.md) - [Scalable GP geostatistics models](./user-guide/geostatistics-models.md) - [CartoBoost Boosting Model Guides](./user-guide/boosting-models/index.md) - [CartoBoost Forecasting Model Guides](./user-guide/forecasting-models/index.md) - [ForecastFrame validation, missing-target, and missing-covariate behavior](./forecasting.md) - [Piecewise linear seasonal forecasting](./user-guide/forecasting-models/piecewise-linear-seasonal.md) - [Naive and seasonal naive forecasting](./user-guide/forecasting-models/naive-seasonal.md) - [Intermittent demand forecasting](./user-guide/forecasting-models/intermittent-demand.md) - [Croston forecasting](./user-guide/forecasting-models/croston.md) - [SBA forecasting](./user-guide/forecasting-models/sba.md) - [TSB forecasting](./user-guide/forecasting-models/tsb.md) - [Graph Spatiotemporal Forecasting, LSTTN Spatial/Temporal Feature Inventory and H3 WASM Debugger, Paper Graph Transformers, and Learned Market Structure](./user-guide/forecasting-models/graph-spatiotemporal.md) - [N-BEATS and N-HiTS neural forecasting](./user-guide/forecasting-models/beats-hits.md) - [Neural panel forecasting](./user-guide/forecasting-models/neural-panel.md) - [Probabilistic and conformal forecasting guide](./user-guide/forecasting-models/probabilistic-conformal.md) - [CartoBoost Graph Model Guides](./user-guide/graph-models/index.md) - [CartoBoost Neural Model Guides](./user-guide/neural-models/index.md) - [Deep model guides](./user-guide/deep-models.md) - [Deep model capability matrix](./reference/capability-matrix.md) - [Deep claim benchmark report](./benchmarks/deep-claims.md) - Deep claim benchmark artifact: `./assets/deep_claim_benchmarks/results.json` - Deep claim benchmark JSONL artifact: `./assets/deep_claim_benchmarks/results.jsonl` - Deep claim benchmark Markdown artifact: `./assets/deep_claim_benchmarks/results.md` - Model capability artifact: `./assets/capabilities/model_capabilities.json` - [Source-only retrieval-augmented diagnostics](./user-guide/deep-models.md) - [Optional foundation model adapters](./user-guide/deep-models.md) - [Source-only multi-view spatial diagnostics](./user-guide/deep-models.md) - [Mixture-of-experts regime modeling](./user-guide/deep-models/cartoboost-regime-moe-forecaster.md) - [Inverted temporal transformer](./user-guide/deep-models/cartoboost-inverted-temporal-transformer.md) - [Delay-aware graph transformer](./user-guide/deep-models/cartoboost-propagation-delay-graph-forecaster.md) - [Conditional flow uncertainty head](./user-guide/deep-models/cartoboost-conditional-flow-distribution-head.md) - [Experimental diffusion scenario generator](./user-guide/deep-models/cartoboost-geotemporal-diffusion-scenario-model.md) - [Advanced experimental neural operators](./user-guide/deep-models/cartoboost-graph-neural-operator.md) - [Choice-set candidate competition](./user-guide/deep-models/cartoboost-choice-set-transformer.md) - [Directional pair forecasting architectures](./user-guide/deep-models/cartoboost-directional-pair-forecaster.md) - [Native deep backend selection and Metal availability](./user-guide/deep-models.md) - [Forecasting reference](./forecasting.md) - [Graph features contract](./graph-features.md) - [Neural features contract](./neural-features.md) - [Sparse H3/S2 and route-cell features](./sparse_features.md) - [Spatial modeling and bearing unit-vector features](./spatial_modeling.md) - [Geo evaluation standard](./user-guide/geo-evaluation-standard.md) - [Geo-causal representation supplements](./user-guide/geo-causal-models.md) - [Benchmarks](./benchmarks/index.md) - [Benchmark methodology and geo manifest contract](./benchmarks/methodology.md) - [NYC Taxi Path C claim gates](./benchmarks/nyc-taxi-path-c.md) - Deep claim benchmark gate: `../scripts/check_deep_claim_gates.py` - Deep claim benchmark harness: `../scripts/run_deep_claim_benchmarks.py` - Capability status gate: `../scripts/check_capability_status.py` - Capability matrix generator: `../scripts/generate_capability_table.py` - Release gate audit: `../scripts/check_release_gates.py` - Public API contract audit: `../scripts/check_public_api_contract.py` - Artifact compatibility audit: `../scripts/check_artifact_compatibility.py` - Docs example audit: `../scripts/check_docs_examples.py` - Official geo evidence audit: `../scripts/check_official_geo_evidence.py` - Performance threshold audit: `../scripts/check_performance_thresholds.py` - AutoGeo is supported-only and excluded from the stable release gate. - NYC TLC quality benchmark harness: `../scripts/run_nyc_taxi_quality_benchmarks.py` - NYC TLC Path C claim harness: `../scripts/run_nyc_taxi_path_c_claims.py` - NYC TLC Path C gate audit: `../scripts/check_nyc_taxi_path_c_gates.py` - [Model benchmark suite](./benchmarks/model-suite.md) - [NYC taxi benchmark report](./benchmarks/nyc-taxi.md) - [Python API reference](./reference/python-api.md) - [CLI reference](./reference/cli.md) ## Model Lists Supported `cartoboost.models.ModelRegistry.defaults()` models: - `models.cartoboost_regressor` / `cartoboost.CartoBoostRegressor` - `models.cartoboost_classifier` / `cartoboost.CartoBoostClassifier` - `models.cartoboost_ranker` / `cartoboost.CartoBoostRanker` - `forecasting.auto_forecaster` / `cartoboost.forecasting.AutoForecaster` - `forecasting.cartoboost_lag` / `cartoboost.forecasting.CartoBoostLagForecaster` - `graph.dcrnn` / `cartoboost.forecasting.DCRNNForecaster` - `geo.nngp` / `cartoboost.geostats.NearestNeighborGPRegressor` - `geo.residual_nngp` / `cartoboost.geostats.ResidualNNGPRegressor` - `geo.spatial_lag` / `cartoboost.spatial_econometrics.SpatialLagRegressor` - `geo.spatial_error` / `cartoboost.spatial_econometrics.SpatialErrorRegressor` - `geo.spatial_durbin` / `cartoboost.spatial_econometrics.SpatialDurbinRegressor` - `causal.synthetic_did` / `cartoboost.geo_causal.SyntheticDIDEstimator` - `causal.geo_lift_design` / `cartoboost.geo_causal.GeoExperimentDesigner` - `causal.spatial_placebo` / `cartoboost.geo_causal.SpatialPlaceboTester` - `prob.conformal_interval` / `cartoboost.prob.ConformalIntervalRegressor` - `prob.spatial_conformal` / `cartoboost.prob.SpatialConformalRegressor` Forecasting model ids and classes: Only `NaiveForecaster`, `SeasonalNaiveForecaster`, `CartoBoostLagForecaster`, `AutoForecaster`, and the typed forecast result/config surfaces are stable in the v0.3 Python API. Other forecasting classes below are supported/demo entries and must be imported from `cartoboost.supported` when available. - `naive` / `NaiveForecaster` - `seasonal_naive` / `SeasonalNaiveForecaster` - `theta` / `ThetaForecaster` - `optimized_theta` / `OptimizedThetaForecaster` - `piecewise_linear_seasonal` / `PiecewiseLinearSeasonalForecaster` - `ets` / `ETSForecaster` - `auto_ets` / Auto ETS browser model - `seasonal_ets` / Seasonal ETS browser model - `auto_arima` / `AutoARIMAForecaster` - `autostats_bank` / `AutoStatsBank` - `classical_expert_bank` / Classical Expert Bank browser model - `croston` / `CrostonForecaster` - `sba` / `SbaForecaster` - `tsb` / `TsbForecaster` - `intermittent_demand` / Intermittent Demand browser model - `kalman` / `KalmanForecaster` - `local_level_kalman` / `LocalLevelKalmanForecaster` - `auto_kalman` / `AutoKalmanForecaster` - `auto_local_level_kalman` / `AutoLocalLevelKalmanForecaster` - `cartoboost_lag` / `CartoBoostLagForecaster` - `cartoboost_direct` / CartoBoost Direct browser model - `rectified_recursive` / Rectified Recursive browser model - `lag_plus` / Lag Plus browser model - `scaled_cartoboost_lag` / Scaled CartoBoost Lag browser model - `log1p_cartoboost_lag` / Log1p CartoBoost Lag browser model - `auto_forecaster` and `auto_forecast` / `AutoForecaster` - `neural_panel` / `NeuralPanelForecaster` - `nbeats` / `NBeatsForecaster` / `NBEATSForecaster` - `nhits` / `NHiTSForecaster` / `NHITSForecaster` - `dcrnn` / `DCRNNForecaster` - `graph_wavenet` / `GraphWaveNetForecaster` - `staeformer` / `STAEformerForecaster` - `stl_cartoboost` / STL + ARIMA browser model - `mstl_cartoboost` / MSTL + ARIMA browser model - `window_average` / Window Average browser model - `seasonal_window_average` / Seasonal Window Average browser model Graph and neural model classes: - `Node2VecStandaloneRegressor` - `GraphSageStandaloneRegressor` - `HeteroGraphSageStandaloneRegressor` - `HinSageStandaloneRegressor` - `Node2VecLinkPredictor` - `GraphSageLinkPredictor` - `HeteroGraphSageLinkPredictor` - `HinSageLinkPredictor` - `NeuralEmbeddingRegressor` - `NeuralEmbeddingStandaloneRegressor` - `NeuralEmbeddingFeatures` - `GraphSageFeatureEncoder` - `HeteroGraphSageFeatureEncoder` - `HinSageFeatureEncoder` - `Node2VecFeatureEncoder` - `GraphFeatureTransformer` Deep model classes and aliases: - `DirectionalPairForecaster` - `ResponseCurveModel` - `EventOutcomeModel` - `ServiceTimeResidualModel` - `SpatioTemporalGraphForecaster` - `ConstrainedDecisionOptimizer` - `TemporalEntityTransformer` - `InvertedTemporalTransformer` - `InvertedEntityTransformer` - `DelayAwareGraphTransformer` - `PropagationDelayGraphForecaster` - `DynamicAdjacencyTransformer` - `RegimeMoEForecaster` - `GeoTemporalMixtureOfExperts` - `PairRegimeRouter` - `EntityRegimeRouter` - `ConditionalFlowDistributionHead` - `JointHorizonFlowHead` - `ResidualFlowCalibrator` - `GeoTemporalDiffusionScenarioModel` - `FlowScenarioGenerator` - `ConditionalResidualDiffusion` - `GraphNeuralOperator` - `FourierGeoOperator` - `SpatioTemporalOperator` - `ChoiceSetTransformer` - `UtilityNet` - `NestedChoiceHead` - `CounterfactualCandidateScorer` Representation, foundation, geo, causal, and probabilistic model surfaces: - `FoundationForecastFeatures`, `ChronosAdapter`, `TimesFMAdapter`, `MoiraiAdapter`, `TimeGPTAdapter`, `TabPFNAdapter`, `TabPFNFeatureGenerator`, `PriorFittedBaseline` - `SpatialLagRegressor`, `SpatialErrorRegressor`, `SpatialDurbinRegressor`, `SpatialTwoStageLeastSquares` - `NearestNeighborGPRegressor`, `ResidualNNGPRegressor`, `SpatialGaussianProcessRegressor` - `SyntheticDIDEstimator`, `GeoLiftEstimator`, `GeoExperimentDesigner`, `SpatialPlaceboTester` - `InvariantRiskEncoder`, `DomainAdversarialGeoEncoder`, `CounterfactualRepresentationNet`, `TreatmentEffectRepresentationHead` - `ConformalIntervalRegressor`, `SpatialConformalRegressor`, `ConformalCalibrator`, `ForecastConformalCalibrator`, `QuantileCartoBoostRegressor` ## What To Emphasize - Lead with the model family, the use case, and the baseline comparison. - Keep implementation details out of the main guides unless they affect how the model is used. - Use taxi examples only when they clarify a structured modeling idea. - [Migrating to CartoBoost 0.3](./migration-v0.3.md)