The HMMSupervisedLearner function takes a vertices table as the input fact table. Each sequence represents a vertex. The PARTITION BY clause consists of list attributes representing the unique sequence across the entire table. The ORDER BY clause sorts the observations in each sequence chronologically in ascending order.
The function can train either one HMM or multiple HMMs. Each model id corresponds to an output HMM model.
| Column | Data Type | Description |
|---|---|---|
| model_column | Any | Identifies the set of observations in a single model. |
| sequence_column | Any | Identifies a sequence of observed values. |
| skip_column | Any data type that can have the value "false", "no", "f", "n", "0", "true", "yes", "t", "y", or "1". | Indicates rows to skip. A value of "true", "yes", "t", "y", or "1" indicates a row to skip. If the value is NULL, the row is not skipped. |
| state_column | Any | Hidden state that generates the observations. |