All Teradata ML Engine analytic functions come preinstalled and ready to use.
All analytic functions take input or output argument clauses. Most argument clauses specify table names, subqueries, or scalar values.
Some analytic functions have argument clauses that specify input or output files. These files typically contain models, dictionaries, or rules required (input) or produced by (output) the function. The default versions of the input files are preinstalled and ready to use.
Each Teradata SQL Engine user has a private schema in Teradata ML Engine for query or function execution. For example, if user alice references a Teradata ML Engine analytic function, she has a private Teradata ML Engine schema named alice. By default, all Teradata ML Engine analytic functions and files are preinstalled in the public schema. Teradata ML Engine system functions (such as available_memory_on_jvm, qginitiatorexport, qginitiatorimport, and so on) are preinstalled in the nc_system schema. Whenever a user, such as alice, references a Teradata ML Engine function or file, the system follows a schema search path to try to find that function or file. For example, if alice runs a KMeans query, the system searches schema alice for a KMeans function, then searches schema public, then nc_system schema.
For a list of analytic functions that use files that are preinstalled on Teradata ML Engine and the files they use, see Teradata® Vantage Machine Learning Engine Analytic Function Reference, B700-4003.
The preinstalled versions of the analytic functions and input files are adequate for most needs. However, you might want to customize the behavior (change defaults) of functions to suit your particular needs. In most cases, you can change function behavior by changing its argument clauses.