Downloads 14this week

GitHub stars 0

Extension repository on GitHub

Extension descriptor (YAML)

Time series forecasting and statistics for DuckDB

Installing and Loading

INSTALL quackstats FROM community;
LOAD quackstats;

Example

SELECT * FROM forecast('sales', timestamp := 'date', value := 'revenue', horizon := 7);

About quackstats

QuackStats brings time series forecasting and statistics directly into DuckDB as SQL table functions.

Forecasting with multiple models (ETS, linear, exponential, logistic, and automatic cross-validation model selection):

SELECT * FROM forecast('sales', timestamp := 'date', value := 'revenue', horizon := 30, model := 'auto');

Grouped forecasting across multiple series in a single query:

SELECT * FROM forecast('sales', timestamp := 'date', value := 'revenue', horizon := 30, group_by := ['region', 'product']);

Seasonality detection to identify periodic patterns:

SELECT * FROM detect_seasonality('sales', timestamp := 'date', value := 'revenue');

All forecasts include configurable prediction intervals (default 95%).

Maintained by the team at Kyomi.

Added Functions