# Teradata AutoARIMA

> Automated ARIMA model selection and forecasting

- Skill: `teradata-labs/teradata-autoarima` (Agent Skill, multi-file: 10 files)
- Install (CLI): `npx skillmds@latest add teradata-labs/teradata-autoarima`
- Raw SKILL.md: https://api.skillmd.com/api/skills/teradata-labs/teradata-autoarima/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Data & Analytics
- Author: teradata-labs (https://skillmd.com/u/teradata-labs)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/teradata-labs/teradata-autoarima

---


# Teradata AutoARIMA

| Property | Value |
|----------|-------|
| **Skill Name** | Teradata AutoARIMA |
| **Description** | Automated ARIMA order selection and parameter estimation |
| **Category** | Time Series Forecasting |
| **Primary Function** | TD_AUTOARIMA |
| **Framework** | UAF (Unbounded Array Framework) |
| **Minimum Version** | Teradata 20.00 |

## Core Capabilities

- Automatic selection of optimal (p,d,q) orders
- Seasonal ARIMA support with automatic period detection
- Stepwise or exhaustive search algorithms
- Chains to TD_ARIMAVALIDATE and TD_ARIMAFORECAST via ART_SPEC
- Fit metrics (AIC, BIC) for model comparison

## Key Parameters

- **MAX_PQ_NONSEASONAL(p,q)**: Maximum non-seasonal AR and MA orders to search
- **STATIONARY(0|1)**: Whether to assume stationarity (skip differencing)
- **STEPWISE(0|1)**: 0=exhaustive search, 1=stepwise (faster)
- **SEASONAL(PERIOD(n))**: Optional seasonal period
- **ALGORITHM(MLE)**: Estimation algorithm
- **FIT_PERCENTAGE(n)**: Percentage of data for fitting (0-100)
- **FIT_METRICS(0|1)**: Return AIC/BIC metrics
- **COEFF_STATS(0|1)**: Return coefficient statistics
- **RESIDUALS(0|1)**: Return model residuals
- **ARMA_ROOTS(0|1)**: Return AR/MA characteristic roots

## Example Usage

```sql
EXECUTE FUNCTION INTO VOLATILE ART(autoarima_results)
TD_AUTOARIMA(
    SERIES_SPEC(
        TABLE_NAME(your_database.your_table),
        ROW_AXIS(TIMECODE(time_column)),
        SERIES_ID(series_id_col),
        PAYLOAD(FIELDS(value_col), CONTENT(REAL))
    ),
    FUNC_PARAMS(
        MAX_PQ_NONSEASONAL(3,3),
        STATIONARY(0),
        STEPWISE(0),
        RESIDUALS(1),
        ARMA_ROOTS(1)
    )
);

SELECT * FROM autoarima_results ORDER BY ROW_I, COL_I;
```

## Use Cases

1. Automated time series forecasting without manual order selection
2. Batch forecasting across many series with varying characteristics
3. Model comparison and selection
4. Seasonal and non-seasonal time series analysis
5. Production forecasting pipelines

## Scripts Included

### Core UAF Scripts
- **`uaf_data_preparation.sql`**: Time series data validation and preparation
- **`td_autoarima_workflow.sql`**: Complete TD_AUTOARIMA implementation
- **`td_autoarima_workflow_template.sql`**: Parameterized workflow template
- **`uaf_pipeline_template.sql`**: Full pipeline with ARIMAVALIDATE and ARIMAFORECAST chaining
- **`uaf_table_analysis.sql`**: Time series structure analysis

### Utility Scripts
- **`parameter_optimization.sql`**: Search space configuration
- **`result_interpretation.sql`**: Model selection results analysis
- **`performance_monitoring.sql`**: UAF execution monitoring

*Teradata AutoARIMA - ClearScape Analytics UAF skill for Teradata Vantage 20.x*