What forecasting models does OctOpus use?

OctOpus rotates across NeuralForecast (NBEATS, PatchTST, xLSTM, TFT), foundation models (Chronos, TiRex, TimesFM, Moirai), and tree-based models with lag and calendar features (LightGBM, XGBoost, CatBoost). The agent picks the family per dataset based on horizon, seasonality, exogenous signals, and series count.

Does OctOpus support hierarchical or multi-series forecasting?

Yes. The agent detects panel structure (per-SKU, per-store, per-customer) and trains a single global model where appropriate. Hierarchical reconciliation is supported when the data has a known hierarchy.

Can OctOpus do zero-shot forecasting with foundation models?

Yes. Chronos, TiRex, TimesFM, and Moirai are tried as zero-shot forecasters when appropriate, then optionally fine-tuned. The agent evaluates each on the same holdout and picks the winner.

Use case · Forecasting

Autonomous AI forecasting — from CSV to deployed prediction API.

Drop your time-series data, describe the horizon, get a forecast. OctOpus picks the right model family across NeuralForecast, Chronos, TiRex, TimesFM, and tree-based models with lags — runs experiments, validates on a real out-of-sample holdout, and deploys a prediction endpoint.

TL;DR. Forecasting is OctOpus's strongest wedge. The agent rotates from a strong tree-based baseline (LightGBM with lags + calendar features) through NeuralForecast (NBEATS, PatchTST, xLSTM, TFT) to foundation models (Chronos, TiRex, TimesFM, Moirai), keeping whichever wins on out-of-sample holdout error.

What OctOpus forecasts well

Demand and inventory — per-SKU, per-store, per-warehouse. Daily, weekly, monthly horizons.
Revenue and KPI — per-customer, per-segment, per-product. Multi-step ahead with confidence bounds.
Energy and load — power consumption, renewable generation, grid balancing.
Sensor and IoT — environmental, industrial, vehicle telemetry.
Hydrology and earth science — rainfall-runoff, river stage, flood horizons. Includes physics-informed feature engineering.
Hierarchical panels — thousands of related series trained as a single global model.

Models the agent rotates through

Tier	Family	When the agent picks it
1 · Baseline	LightGBM with lag, rolling, calendar features	Almost always — fast, strong, interpretable.
2 · Tuned GBM	XGBoost / CatBoost with Optuna	When tier 1 has headroom and the data justifies a search.
3 · Deep / modern	NeuralForecast — NBEATS, PatchTST, xLSTM, TFT	Long horizons, multi-series panels, exogenous signals.
4 · Foundation	Chronos, TiRex, TimesFM, Moirai	Zero-shot first; cold-start data; benchmark anchor.
5 · Stacking	Linear / GBM stacker over diverse base learners	When the residuals of different families are uncorrelated.

How a forecasting run looks

Profile. The agent reads the schema, detects the time column, infers granularity (daily / hourly / weekly), spots multiple series, and flags exogenous candidates.
Plan. Writes a research spec: target, horizon, validation strategy (rolling-origin or fixed cutoff), and which families to try in which order.
Run. Generates a fresh train.py per experiment, executes in a sandbox, reads the metrics.
Diagnose. When something fails, the agent reads the traceback, writes a targeted fix (dtype, missing dependency, cardinality blow-up), and retries.
Validate. The winner is scored on a holdout slice the LLM never sees.
Deploy. A prediction endpoint plus a downloadable bundle (train.py, model.pkl, deploy.zip).

What you get back

Forecast with confidence intervals at the requested horizon.
Holdout error per series and aggregate (sMAPE, MAPE, MASE, MAE — the agent picks per data shape).
Calibration plot and residual diagnostics.
The exact train.py the agent wrote — fully inspectable, fully reproducible.
A deployable prediction API.

Forecast your data free → See benchmarks vs traditional AutoML

Autonomous AI forecasting — from CSV to deployed prediction API.

What OctOpus forecasts well

Models the agent rotates through

How a forecasting run looks

What you get back

Related use cases