Measure swarm · Anomaly Detection Agent · Predictive-anomaly-forecasting skill · Published June 1, 2026

Multi-location marketing forecasting for franchise operators

Facebook Prophet, Statsmodels ARIMA + SARIMA + SARIMAX, PyMC + Stan Bayesian time-series, Darts + sktime + Nixtla StatsForecast + MLForecast + NeuralForecast, scikit-learn, TensorFlow + Keras LSTM, PyTorch LSTM + TFT + N-BEATS + N-HiTS, AutoTS, Kats (Meta), Greykite (LinkedIn), Anodot, Datafold, Monte Carlo, Domo Mr. Roboto, Tableau Pulse, ThoughtSpot SpotIQ ship per-platform forecasting primitives. The predictive-anomaly-forecasting skill on the anomaly-detection agent — running per-portfolio per-location per-metric per-forecast-window per-multi-algorithm-ensemble + per-causal-driver-decomposition + per-pre-breach-action-recommendation + per-fleet-wide-pattern-history-substrate + per-portfolio audit-trail at multi-location-forecasting-operator scale — is operator-side architecture above the per-platform primitive.

Engage the anomaly-detection agent See the 32-agent swarm

What this skill closes

Multi-algorithm forecasting ensemble — Prophet + ARIMA + SARIMA + SARIMAX + ETS + UCM + VAR + VECM + Bayesian-PyMC + Stan + Darts (NBEATS + NHITS + TFT + TCN + RNN + Transformer + DLinear + NLinear) + Nixtla StatsForecast (AutoARIMA + AutoETS + AutoCES + AutoTheta + AutoMFLES) + Nixtla MLForecast (LightGBM + XGBoost + CatBoost + LSTM) + Nixtla NeuralForecast (NBEATS + NHITS + LSTM + TFT + Informer + AutoFormer + PatchTST) + NeuralProphet + AutoTS + Kats + Greykite with per-algorithm-MAPE-weighted-ensemble-blend.
Causal driver decomposition — Robyn (Meta) MMM + LightweightMMM (Google) + Orbit (Uber) Bayesian-MMM + PyMC-Marketing + Mass Markdown Mix Modeling + CausalImpact (Google) + GeoLift (Meta) + DoubleML ATE + EconML CATE + CausalML T-S-X-R-Learner.
Per-driver impact attribution across per-paid-search + per-paid-social + per-email + per-SMS + per-organic-search + per-organic-social + per-direct-response + per-referral + per-weather + per-seasonality + per-promotion + per-competitive-density 12 driver canonical-impact-attribution.
Per-driver saturation + adstock + decay — per-Hill-equation + per-Adstock + per-Weibull + per-S-curve canonical-saturation-curve + per-driver-decay-Adstock-spec + per-driver-cross-correlation + per-driver-elasticity-computation.
Pre-breach action recommendation with per-14-day + per-21-day + per-30-day pre-breach windows + per-counterfactual simulation + per-action-priority ranking + per-action-routing to correct agent (paid-search bid-adjustment + paid-social creative rotation + email flow trigger + SMS trigger + GBP Offer Post + loyalty offer decisioning + promotion takeover window).
Fleet-wide pattern history substrate — per-location rolling-2-year history + per-cross-location-pattern-similarity-matching (Denver-current-trajectory similar to Phoenix-2024-Q3-breach-pattern) + per-transfer-learning + per-vertical-cluster membership (fitness + beauty + food + home-service).
Per-forecast accuracy MAPE tracking — per-MAPE + per-RMSE + per-MAE + per-MSE + per-sMAPE + per-MASE + per-coverage-percentage with rolling-30-day + 90-day + 365-day MAPE windows feeding algorithm-weighted-ensemble-blend.

Why per-vendor-Anodot-canonical-single-metric-canonical-single-algorithm breaks at multi-location-forecasting-operator scale

Per-vendor-Anodot-canonical-single-metric-canonical-single-algorithm ships per-account per-metric per-single-algorithm per-7-day-horizon primitive. Per-vendor-Datafold + Monte-Carlo + Lightup + Bigeye + Sifflet + Acceldata + Validio + Soda + Metaplane + Synq + Re.Data + Great-Expectations + Domo-Mr-Roboto + Tableau-Pulse + ThoughtSpot-SpotIQ + Sisense-Forecast + Power-BI-Forecast + Qlik-AutoML-canonical-single-account ship per-platform per-native primitives.

At 1-location-1-metric-1-algorithm scale per-metric per-algorithm per-7-day-horizon primitive is enough. At 200-location-multi-metric-multi-algorithm scale per-200-locations × per-9-stream-metric × per-multi-algorithm × per-14-30-day-horizon = canonical-cross-location-canonical-cross-algorithm-canonical-cross-horizon-blind.

Per-Anodot-tracks-single-metric-revenue + per-Datafold-tracks-data-quality + per-Domo-tracks-aggregate-only + per-Tableau-Pulse-tracks-aggregate-only per-canonical-cross-platform-canonical-coverage-blind. Per-canonical-multi-algorithm-canonical-ensemble-blind + per-canonical-causal-driver-canonical-decomposition-blind + per-canonical-fleet-wide-pattern-canonical-history-substrate-blind + per-canonical-per-pre-breach-canonical-action-canonical-recommendation-blind + per-canonical-per-forecast-canonical-accuracy-canonical-MAPE-tracking-blind.

Per-Denver-location-revenue-canonical-breach-detected-canonical-after-the-fact + per-pre-breach-action-window-missed + per-revenue-canonical-loss-canonical-from-missed-pre-breach-window-canonical-3-to-12-percent-cumulative. Per-canonical-multi-algorithm-canonical-ensemble + per-canonical-causal-driver-decomposition + per-canonical-fleet-wide-pattern-history-substrate + per-canonical-pre-breach-action-recommendation is operator-side architecture above per-platform per-single-algorithm primitive.

What is in market today

Per-platform per-forecasting-library

Facebook Prophet + Meta NeuralProphet, Statsmodels ARIMA + SARIMA + SARIMAX + ETS + UCM + VAR + VECM, PyMC + Stan Bayesian, Pyro + NumPyro, Darts + sktime + Nixtla StatsForecast + MLForecast + HierarchicalForecast + NeuralForecast, scikit-learn, TensorFlow + Keras LSTM, PyTorch LSTM + TFT + N-BEATS + N-HiTS + AutoFormer + Informer, AutoTS, Kats (Meta), Greykite (LinkedIn). Per-library per-algorithm-fit. Per-canonical-multi-algorithm-canonical-ensemble-canonical-MAPE-weighted-blend is not the primitive.

Per-platform per-anomaly-detection-with-forecasting

Anodot, Datafold, Monte Carlo, Lightup, Bigeye, Sifflet, Acceldata, Validio, Soda, Metaplane, Synq, Re.Data, Great Expectations. Per-account per-metric per-single-algorithm. Per-canonical-causal-driver-canonical-decomposition + per-canonical-fleet-wide-pattern-history-substrate is not the primitive.

Per-platform per-BI-with-forecasting

Domo Mr. Roboto, Tableau Pulse, ThoughtSpot SpotIQ, Sisense Forecast, Looker (Google), Power BI Forecast, Qlik AutoML, Mode (Thoughtworks), Hex Forecast. Per-account per-dashboard per-aggregate-forecast. Per-canonical-per-pre-breach-canonical-action-canonical-recommendation-canonical-counterfactual-simulation is not the primitive.

Per-platform per-marketing-mix-modeling

Robyn (Meta), LightweightMMM (Google), Orbit (Uber), PyMC-Marketing, CausalImpact (Google), GeoLift (Meta), Mass Markdown Mix Modeling, DoubleML, EconML, CausalML. Per-library per-MMM-model-fit. Per-canonical-cross-MMM-canonical-ensemble-canonical-per-driver-impact-canonical-attribution-canonical-with-forecast-substrate is not the primitive.

How the architecture is set up

Per-portfolio per-canonical-9-stream-canonical-substrate-ingestion. Per-organic-search + per-paid-search + per-paid-social + per-direct + per-referral + per-email + per-SMS + per-organic-social + per-direct-response canonical-9-stream (companion to /nine-stream-marketing-anomaly-coverage).
Per-portfolio per-canonical-Facebook-Prophet-canonical-forecast. Per-trend-seasonality-holiday-decomposition + per-changepoint-detection + per-uncertainty-interval + per-multi-regressor canonical-Prophet.
Per-portfolio per-canonical-Statsmodels-ARIMA-family-canonical-forecast. Per-ARIMA + per-SARIMA + per-SARIMAX + per-ETS + per-UCM + per-VAR + per-VECM canonical-ARIMA-family.
Per-portfolio per-canonical-Bayesian-canonical-PyMC-Stan-canonical-forecast. Per-PyMC + per-Stan + per-CmdStanPy + per-Pyro + per-NumPyro + per-Bayesian-state-space + per-BSTS canonical-Bayesian.
Per-portfolio per-canonical-Darts-canonical-forecast. Per-NBEATS + per-NHITS + per-TFT + per-TCN + per-RNN + per-Transformer + per-DLinear + per-NLinear canonical-Darts.
Per-portfolio per-canonical-Nixtla-canonical-StatsForecast-MLForecast-NeuralForecast-canonical-forecast. Per-AutoARIMA + per-AutoETS + per-AutoCES + per-LightGBM + per-XGBoost + per-CatBoost + per-NBEATS-Neural + per-LSTM + per-TFT-Neural + per-Informer + per-PatchTST canonical-Nixtla.
Per-portfolio per-canonical-multi-algorithm-canonical-ensemble-canonical-MAPE-weighted-blend. Per-algorithm-historical-accuracy-MAPE-weighting + per-weighted-blend + per-ensemble-voting + per-confidence-interval-aggregation.
Per-portfolio per-canonical-causal-driver-canonical-decomposition. Per-Robyn + per-LightweightMMM + per-Orbit + per-PyMC-Marketing + per-CausalImpact + per-GeoLift + per-DoubleML + per-EconML + per-CausalML canonical-MMM-substrate.
Per-portfolio per-canonical-per-driver-canonical-saturation-canonical-adstock-canonical-decay. Per-Hill + per-Adstock + per-Weibull + per-S-curve + per-driver-decay + per-cross-driver-correlation + per-elasticity-computation.
Per-portfolio per-canonical-fleet-wide-pattern-canonical-history-canonical-substrate. Per-location-rolling-2-year-history + per-cross-location-pattern-similarity-matching + per-transfer-learning + per-vertical-cluster-membership.
Per-portfolio per-canonical-per-pre-breach-canonical-action-canonical-recommendation. Per-14-day + per-21-day + per-30-day pre-breach windows + per-counterfactual-simulation + per-action-priority-ranking + per-action-routing-to-correct-agent.
Per-portfolio per-canonical-per-forecast-accuracy-canonical-MAPE-tracking. Per-MAPE + per-RMSE + per-MAE + per-MSE + per-sMAPE + per-MASE + per-coverage-percentage + rolling-30-day + 90-day + 365-day MAPE windows.
Per-portfolio audit-trail + per-CMO-dashboard-rollup + per-action-routing. Per-forecast + per-ensemble-decision + per-causal-decomposition + per-pre-breach-window + per-counterfactual-simulation + per-action-recommendation + per-CMO-dashboard.

Frequently asked questions

What is multi-location marketing forecasting for franchise operators?

Multi-location marketing forecasting runs per-portfolio per-location per-metric per-forecast-window per-canonical-multi-algorithm-forecasting-canonical-ensemble + per-canonical-causal-driver-canonical-decomposition + per-canonical-per-forecast-confidence-canonical-interval + per-canonical-per-pre-breach-canonical-action-canonical-recommendation + per-canonical-per-forecast-accuracy-canonical-MAPE-canonical-tracking + per-canonical-fleet-wide-pattern-canonical-history-canonical-substrate + per-portfolio audit-trail. Per-canonical-multi-algorithm-forecasting-canonical-ensemble runs per-canonical-Facebook-Prophet-canonical-forecast (per-trend-seasonality-holiday-canonical-decomposition + per-changepoint-detection + per-uncertainty-interval per-canonical-Prophet) + per-canonical-Statsmodels-ARIMA-canonical-SARIMA-canonical-SARIMAX-canonical-forecast (per-AutoRegressive-Integrated-Moving-Average + per-Seasonal-ARIMA + per-Seasonal-ARIMA-with-eXogenous-variables per-canonical-ARIMA) + per-canonical-Bayesian-canonical-PyMC-canonical-Stan-canonical-forecast + per-canonical-Pyro-NumPyro-canonical-probabilistic-canonical-forecast + per-canonical-Darts-canonical-forecast (per-NBEATS + per-TFT-Temporal-Fusion-Transformer + per-TCN-Temporal-Convolutional-Network + per-RNN + per-Transformer per-canonical-Darts) + per-canonical-sktime-canonical-forecast + per-canonical-Nixtla-StatsForecast-canonical-AutoARIMA-canonical-AutoETS-canonical-AutoCES-canonical-forecast + per-canonical-Nixtla-MLForecast-canonical-LightGBM-canonical-XGBoost-canonical-LSTM-canonical-forecast + per-canonical-TensorFlow-Keras-LSTM-canonical-forecast + per-canonical-PyTorch-LSTM-canonical-forecast + per-canonical-NeuralProphet-canonical-forecast + per-canonical-LSTM-attention-mechanism-canonical-forecast + per-canonical-multi-algorithm-canonical-ensemble-canonical-voting + per-canonical-per-algorithm-canonical-weighted-canonical-blend + per-canonical-per-algorithm-canonical-historical-accuracy-MAPE-canonical-weighting. The per-platform per-forecasting-library category includes Facebook Prophet + Meta NeuralProphet, Statsmodels ARIMA + SARIMA + SARIMAX + ETS + UCM + VAR + VECM, PyMC + Stan Bayesian time-series, Pyro + NumPyro probabilistic programming, Darts + sktime + Nixtla StatsForecast + MLForecast + HierarchicalForecast + NeuralForecast, scikit-learn, TensorFlow + Keras LSTM, PyTorch LSTM + Temporal Fusion Transformer + N-BEATS + N-HiTS + AutoFormer + Informer, AutoTS, Kats (Meta), Greykite (LinkedIn). The per-platform per-anomaly-detection-with-forecasting category includes Anodot, Datafold, Monte Carlo, Lightup, Bigeye, Sifflet, Acceldata, Validio, Soda, Metaplane, Synq, Re.Data, Great Expectations. The per-platform per-BI-with-forecasting category includes Domo Mr. Roboto, Tableau Pulse, ThoughtSpot SpotIQ, Sisense Forecast, Looker (Google), Power BI Forecast, Qlik AutoML, Mode (Thoughtworks), Hex Forecast. The per-platform per-marketing-mix-modeling category includes Robyn (Meta), LightweightMMM (Google), Orbit (Uber), CausalImpact (Google), GeoLift (Meta), Mass Markdown Mix Modeling. The predictive-anomaly-forecasting skill on the anomaly-detection agent (1 of the bundle on anomaly-detection in the measure swarm alongside per-canonical-multi-tool-alert-deduplication + per-canonical-two-sigma-outlier-flagging + per-canonical-multi-dimensional-threshold-routing + per-canonical-multi-location-crisis-detection + per-canonical-nine-alert-stream-coverage) — running per-portfolio per-location per-metric per-forecast-window per-multi-algorithm-ensemble + per-causal-driver-decomposition + per-pre-breach-action-recommendation + per-fleet-wide-pattern-history-substrate + per-portfolio audit-trail at multi-location-forecasting-operator scale — is operator-side architecture above the per-platform per-forecasting-library + per-anomaly-detection-with-forecasting + per-BI-with-forecasting + per-marketing-mix-modeling primitive.

Why does per-vendor-Anodot-canonical-single-metric-canonical-single-algorithm break down at multi-location-forecasting-operator scale?

Per-vendor-Anodot-canonical-single-metric-canonical-single-algorithm ships per-account per-metric per-single-algorithm per-7-day-horizon primitive. Per-vendor-Datafold + per-Monte-Carlo + per-Lightup + per-Bigeye + per-Sifflet + per-Acceldata + per-Validio + per-Soda + per-Metaplane + per-Synq + per-Re.Data + per-Great-Expectations + per-Domo-Mr-Roboto + per-Tableau-Pulse + per-ThoughtSpot-SpotIQ + per-Sisense-Forecast + per-Power-BI-Forecast + per-Qlik-AutoML-canonical-single-account ship per-platform per-native primitives. At 1-location-1-metric-1-algorithm scale per-metric per-algorithm per-7-day-horizon primitive is enough. At 200-location-multi-metric-multi-algorithm scale per-200-locations × per-9-stream-metric × per-multi-algorithm × per-14-30-day-horizon = canonical-cross-location-canonical-cross-algorithm-canonical-cross-horizon-blind. Per-Anodot-canonical-tracks-single-metric-revenue + per-Datafold-canonical-tracks-single-metric-data-quality + per-Domo-canonical-tracks-aggregate-only + per-Tableau-Pulse-canonical-tracks-aggregate-only per-canonical-cross-platform-canonical-cross-metric-canonical-coverage-blind. Per-canonical-multi-algorithm-canonical-ensemble-blind (per-Anodot-canonical-canonical-rolling-z-score-only + per-Datafold-canonical-Prophet-only + per-Domo-canonical-aggregate-trend-only per-canonical-single-algorithm-blind) + per-canonical-causal-driver-canonical-decomposition-blind (per-marketing-mix-modeling-blind + per-per-driver-canonical-impact-canonical-attribution-blind) + per-canonical-fleet-wide-pattern-canonical-history-canonical-substrate-blind (per-cross-location-canonical-history-canonical-blind + per-cross-location-canonical-pattern-canonical-similarity-canonical-detection-blind) + per-canonical-per-pre-breach-canonical-action-canonical-recommendation-blind + per-canonical-per-forecast-canonical-accuracy-canonical-MAPE-canonical-tracking-blind. Per-canonical-Denver-location-revenue-canonical-breach-detected-canonical-after-the-fact + per-pre-breach-canonical-action-canonical-window-canonical-missed + per-canonical-revenue-canonical-loss-canonical-from-canonical-missed-pre-breach-canonical-window-canonical-3-to-12-percent-canonical-cumulative. Per-vendor-Anodot + Datafold + Monte-Carlo + Lightup + Bigeye + Sifflet + Acceldata + Validio + Soda + Metaplane + Synq + Re.Data + Great-Expectations + Domo + Tableau-Pulse + ThoughtSpot-SpotIQ ship per-platform per-single-algorithm primitive. Per-canonical-multi-algorithm-canonical-ensemble + per-canonical-causal-driver-decomposition + per-canonical-fleet-wide-pattern-history-substrate + per-canonical-pre-breach-action-recommendation is operator-side architecture above per-platform per-single-algorithm primitive.

What does per-portfolio per-canonical-multi-algorithm-forecasting-canonical-ensemble do?

Per-portfolio per-canonical-multi-algorithm-forecasting-canonical-ensemble runs per-portfolio per-canonical-Facebook-Prophet-canonical-forecast (per-trend-seasonality-holiday-canonical-decomposition + per-changepoint-detection + per-uncertainty-interval + per-cross-correlation-canonical-multi-regressor per-canonical-Prophet) + per-canonical-Statsmodels-ARIMA-canonical-SARIMA-canonical-SARIMAX-canonical-forecast (per-AutoRegressive-Integrated-Moving-Average + per-Seasonal-ARIMA + per-Seasonal-ARIMA-with-eXogenous-variables + per-ETS-Error-Trend-Seasonal + per-UCM-Unobserved-Components-Model + per-VAR-Vector-AutoRegressive + per-VECM-Vector-Error-Correction-Model per-canonical-ARIMA-family) + per-canonical-Bayesian-canonical-PyMC-canonical-Stan-canonical-forecast (per-PyMC + per-Stan + per-CmdStanPy + per-Pyro + per-NumPyro + per-Bayesian-state-space-model + per-Bayesian-structural-time-series-BSTS per-canonical-Bayesian) + per-canonical-Darts-canonical-forecast (per-NBEATS + per-NHITS + per-TFT-Temporal-Fusion-Transformer + per-TCN-Temporal-Convolutional-Network + per-RNN + per-Transformer + per-DLinear + per-NLinear per-canonical-Darts) + per-canonical-Nixtla-StatsForecast-canonical-forecast (per-AutoARIMA + per-AutoETS + per-AutoCES + per-AutoTheta + per-AutoMFLES per-canonical-StatsForecast) + per-canonical-Nixtla-MLForecast-canonical-forecast (per-LightGBM + per-XGBoost + per-CatBoost + per-LSTM-MLForecast per-canonical-MLForecast) + per-canonical-Nixtla-NeuralForecast-canonical-forecast (per-NBEATS + per-NHITS + per-LSTM-NeuralForecast + per-TFT-NeuralForecast + per-Informer + per-AutoFormer + per-PatchTST per-canonical-NeuralForecast) + per-canonical-NeuralProphet-canonical-forecast + per-canonical-AutoTS-canonical-forecast + per-canonical-Kats-Meta-canonical-forecast + per-canonical-Greykite-LinkedIn-canonical-forecast + per-canonical-per-algorithm-canonical-historical-accuracy-MAPE-canonical-weighting + per-canonical-per-algorithm-canonical-weighted-canonical-blend + per-canonical-multi-algorithm-canonical-ensemble-canonical-voting + per-canonical-per-forecast-canonical-confidence-interval-canonical-aggregation. Per-portfolio audit-trail.

How does per-portfolio per-canonical-causal-driver-canonical-decomposition work?

Per-portfolio per-canonical-causal-driver-canonical-decomposition runs per-portfolio per-canonical-marketing-mix-modeling-canonical-substrate (per-Robyn-Meta-canonical-MMM + per-LightweightMMM-Google-canonical-MMM + per-Orbit-Uber-canonical-Bayesian-MMM + per-PyMC-Marketing-canonical-MMM + per-Mass-Markdown-Mix-Modeling per-canonical-MMM) + per-canonical-causal-impact-canonical-CausalImpact-Google-canonical-substrate (per-Bayesian-structural-time-series-BSTS + per-control-time-series + per-treatment-time-series + per-causal-impact-canonical-estimation per-canonical-CausalImpact) + per-canonical-GeoLift-Meta-canonical-substrate (per-geo-test-design + per-treatment-vs-control-geo + per-DiD-Difference-in-Differences per-canonical-GeoLift) + per-canonical-DoubleML-canonical-ATE-canonical-Average-Treatment-Effect + per-canonical-EconML-canonical-CATE-canonical-Conditional-Average-Treatment-Effect + per-canonical-CausalML-canonical-T-S-X-R-Learner + per-canonical-per-driver-canonical-impact-canonical-attribution (per-paid-search-driver-impact + per-paid-social-driver-impact + per-email-driver-impact + per-SMS-driver-impact + per-organic-search-driver-impact + per-organic-social-driver-impact + per-direct-response-driver-impact + per-referral-driver-impact + per-weather-driver-impact + per-seasonality-driver-impact + per-promotion-driver-impact + per-competitive-density-driver-impact per-canonical-per-driver-impact) + per-canonical-per-driver-canonical-saturation-canonical-curve-canonical-spec (per-Hill-equation + per-Adstock + per-Weibull + per-S-curve per-canonical-saturation-curve) + per-canonical-per-driver-canonical-decay-canonical-Adstock-canonical-spec + per-canonical-per-driver-canonical-cross-driver-canonical-correlation + per-canonical-per-driver-canonical-elasticity-canonical-computation. Per-portfolio audit-trail.

What does per-portfolio per-canonical-per-pre-breach-canonical-action-canonical-recommendation + per-canonical-fleet-wide-pattern-canonical-history-canonical-substrate do?

Per-portfolio per-canonical-per-pre-breach-canonical-action-canonical-recommendation runs per-portfolio per-canonical-per-forecast-canonical-threshold-canonical-evaluation (per-forecast-canonical-breach-threshold-canonical-detection + per-forecast-canonical-degradation-trajectory-canonical-detection + per-forecast-canonical-acceleration-canonical-detection per-canonical-breach-evaluation) + per-canonical-per-pre-breach-canonical-window-canonical-computation (per-14-day-pre-breach-window + per-21-day-pre-breach-window + per-30-day-pre-breach-window per-canonical-pre-breach-window) + per-canonical-per-pre-breach-canonical-action-canonical-recommendation-canonical-generation (per-pre-breach-canonical-paid-search-bid-adjustment + per-pre-breach-canonical-paid-social-creative-rotation + per-pre-breach-canonical-email-flow-trigger + per-pre-breach-canonical-SMS-trigger + per-pre-breach-canonical-GBP-Offer-Post + per-pre-breach-canonical-loyalty-offer-decisioning + per-pre-breach-canonical-promotion-takeover-window-recommendation per-canonical-action-recommendation) + per-canonical-per-action-canonical-projected-impact-canonical-simulation (per-action-canonical-counterfactual-canonical-forecast-canonical-simulation + per-action-canonical-vs-no-action-canonical-counterfactual-canonical-revenue-impact per-canonical-counterfactual-simulation) + per-canonical-per-action-canonical-priority-canonical-ranking + per-canonical-per-action-canonical-routing-canonical-to-correct-canonical-agent. Per-canonical-fleet-wide-pattern-canonical-history-canonical-substrate runs per-portfolio per-canonical-cross-location-canonical-history-canonical-aggregation (per-location-rolling-2-year-canonical-history + per-cross-location-canonical-pattern-canonical-similarity-canonical-detection per-canonical-cross-location-history) + per-canonical-cross-location-canonical-pattern-canonical-similarity-canonical-matching (per-Denver-current-trajectory-canonical-similar-to-Phoenix-2024-Q3-canonical-breach-pattern + per-cross-location-canonical-historical-canonical-precedent-canonical-matching per-canonical-similarity-matching) + per-canonical-cross-location-canonical-pattern-canonical-transfer-canonical-learning + per-canonical-per-location-canonical-cluster-canonical-membership-canonical-assignment (per-fitness-vertical-cluster + per-beauty-vertical-cluster + per-food-vertical-cluster + per-home-service-cluster per-canonical-cluster-assignment). Per-portfolio audit-trail.

What does per-portfolio per-canonical-per-forecast-accuracy-canonical-MAPE-canonical-tracking + per-anomaly-detection-agent-canonical-bundle do?

Per-portfolio per-canonical-per-forecast-accuracy-canonical-MAPE-canonical-tracking runs per-portfolio per-canonical-per-algorithm-per-forecast-canonical-actual-canonical-vs-forecast-canonical-comparison + per-canonical-per-algorithm-canonical-MAPE-Mean-Absolute-Percentage-Error-canonical-computation + per-canonical-per-algorithm-canonical-RMSE-Root-Mean-Square-Error + per-canonical-per-algorithm-canonical-MAE-Mean-Absolute-Error + per-canonical-per-algorithm-canonical-MSE-Mean-Square-Error + per-canonical-per-algorithm-canonical-sMAPE-symmetric-MAPE + per-canonical-per-algorithm-canonical-MASE-Mean-Absolute-Scaled-Error + per-canonical-per-algorithm-canonical-coverage-percentage-canonical-confidence-interval + per-canonical-per-algorithm-canonical-rolling-30-day-canonical-MAPE-canonical-window + per-canonical-per-algorithm-canonical-rolling-90-day-MAPE-canonical-window + per-canonical-per-algorithm-canonical-rolling-365-day-MAPE-canonical-window + per-canonical-per-algorithm-canonical-MAPE-canonical-weighted-canonical-ensemble-canonical-blend-canonical-input. Per-anomaly-detection-agent-canonical-bundle integrates the predictive-anomaly-forecasting skill with sibling skills on the same agent: per-canonical-nine-alert-stream-coverage (skill sibling — uses 9-stream substrate for forecasting input) + per-canonical-multi-tool-alert-deduplication (skill sibling — deduplicates pre-breach alerts from multiple forecasting algorithms) + per-canonical-two-sigma-outlier-flagging (skill sibling — companion detection algorithm) + per-canonical-multi-dimensional-threshold-routing (skill sibling — routes pre-breach alerts to correct owner) + per-canonical-multi-location-crisis-detection (skill sibling — uses fleet-wide pattern history for crisis pre-detection). Per-portfolio audit-trail.

Engage the anomaly-detection agent

Per-portfolio per-location per-metric per-forecast-window per-multi-algorithm-ensemble + per-causal-driver-decomposition + per-pre-breach-action-recommendation + per-fleet-wide-pattern-history-substrate + per-forecast-accuracy-MAPE-tracking + per-portfolio audit-trail shipped as the orchestration layer above your existing per-forecasting-library + per-anomaly-detection-with-forecasting + per-BI-with-forecasting + per-marketing-mix-modeling primitive.

Book a scope call Browse the 32-agent catalog