The DTW function requires three input tables:
| Input Table | Description |
|---|---|
| input_table | Each row contains information for one time series. |
| template_table | Each row contains information for one time series. |
| mapping_table | Defines the mapping between input_table rows and template_table rows. |
The columns by which input_table and mapping_table are partitioned must agree in number and data type. That is, each i_partition_column must have a corresponding m_partition_column with the same data type. However, corresponding partition columns can have different names.
The columns by which input_table and template_table are ordered must agree in number and data type. That is, each i_ordering_column must have a corresponding t_ordering_column with the same data type. However, corresponding ordering columns can have different names.
| Column Name | Data Type | Description |
|---|---|---|
| timeseriesid | INTEGER | Unique identifier for a time series. |
| timestamp | INTEGER, SMALLINT, BIGINT, NUMERIC, or DOUBLE PRECISION | Timestamp for the time series. |
| value | DOUBLE PRECISION | Value for the timestamp. |
| Column Name | Data Type | Description |
|---|---|---|
| templateid | INTEGER | Unique identifier for a time series. |
| timestamp | INTEGER, SMALLINT, BIGINT, NUMERIC, or DOUBLE PRECISION | Timestamp for the time series. |
| value | DOUBLE PRECISION | Value for the timestamp. |
| Column Name | Data Type | Description |
|---|---|---|
| timeseriesid | INTEGER | Unique identifier for a time series in input_table. |
| templateid | INTEGER | Unique identifier for the time series in template_table that corresponds to the time series specified by timeseriesid. |
In mapping_table, DTW supports a single ID column in the input_table and template_table.