Template-based Time Series Generation with Loom

Time series analysis and forecasting are important techniques for decision-making in many domains. They are typically evaluated on given sets of time series that have a constant size and specified characteristics. Synthetic datasets are relevant because they are flexible in both size and characteristics. In this demo, we present our prototype Loom, that generates datasets with respect to the user’s configuration of categorical information and time series characteristics. The prototype allows for comparison of different analysis techniques.