12.7.4 Special Control Arguments (Autonomous Database)

Use the PAR_LST parameter to specify special control arguments and additional arguments to be passed into the Python script.

Argument	Syntax and Description
oml_input_type	Syntax oml_input_type : 'pandas.DataFrame', 'numpy.recarray', or 'default' (default) Description Specifies the type of object to construct from data in the Autonomous Database. By default, a two-dimensional numpy.ndarray of type numpy.float64 is constructed if all columns are numeric. Otherwise, a pandas.DataFrame is constructed.
oml_na_omit	Syntax oml_na_omit : bool, false (default) Description Determines if rows with any missing values will be omitted from the table to be evaluated. If true, omit all rows with missing values from the table. If false, do not omit rows with missing values from the table.
oml_async_flag	Syntax oml_async_flag: bool, false (default) Description If true, the job will be submitted asynchronously. If false, the job will be executed in synchronous mode.
oml_graphics_flag	Syntax oml_graphics_flag: bool, false (default) Description If true, the server will capture images rendered in the Python script. If false, the server will not capture images rendered in the Python script.
oml_parallel_flag	Syntax oml_parallel_flag: bool, false (default) Description If true, the Python script will be run with data parallelism. Data parallelism is only applicable to pyqRowEval, pyqGroupEval, and pyqIndexEval. If false, the Python script will not be run with data parallelism.
oml_service_level	Syntax oml_service_level : string, allowed values: 'LOW'(default), 'MEDIUM', 'HIGH' Description Controls the different levels of performance and concurrency in Autonomous Database.

Examples

• Input data is pandas.DataFrame:

  par_lst => '{"oml_input_type":"pandas.DataFrame"}'

• Drop rows with missing values from input data:

  par_lst => '{"oml_na_omit":true}'

• Submit a job in asynchronous mode:

  par_lst => '{"oml_async_flag":true}'

• Use MEDIUM service level:

  par_lst => '{"oml_service_level":"MEDIUM"}'