Predefined Spatial Functions Available from OML Embedded Python Execution

Spatial AI provides some pre-defined spatial functions. You can call the register_sai_scripts function (in oraclesai.oml) to register the pre-defined spatial functions for embedded execution.

The following code registers the pre-defined spatial functions for embedded execution into the script repository and lists them using the oml.script.dir function.

import oml
from  oraclesai.oml import register_sai_scripts
 
# register for all the users and overwrite if already registered
register_sai_scripts(is_global=True, overwrite=True)
 
# list registered scripts
oml.script.dir(sctype='all')[['name']]

Note the errorModelPredict function (defined in Store a Function for Embedded Execution) in the list of registered functions along with the other pre-defined functions.

                                     name
0                              clustering
1  compute_global_spatial_autocorrelation
2   compute_local_spatial_autocorrelation
3                 compute_spatial_weights
4                      create_spatial_lag
5                       errorModelPredict

The following table lists the parameters that are required for all the pre-defined spatial functions.

Parameter	Description
oml_connect	This parameter must always be true since all the pre-defined spatial functions require a connection to the database.
table	The name of a database table.

The rest of the parameters may vary depending on the spatial function.

The following table describes each one of the pre-defined spatial functions for embedded execution.

Spatial Functions	Description
compute_spatial_weights	This function computes the spatial weights for the given spatial table and stores a SpatialWeights object in the data store according to the save_weights_as parameter.
compute_global_spatial_autocorrelation	Computes the Moran’s I statistic for the given spatial table and column. The function returns the value of the Moran’s I statistic, its z-value, and its p-value.
compute_local_spatial_autocorrelation	Calculates the Local Moran’s I statistic of all the observations from the given spatial table. The function returns a table containing the Local Moran’s I statistic for each row, along with the z-value and p-value.
create_spatial_lag	Computes the spatial lag for the given column of the provided spatial table. The function returns a table with the calculated spatial lag for each row from the input table.
clustering	Executes a clustering algorithm with the data from the given spatial table, using only the specified columns or all the columns if the columns parameter is not provided. Available clustering methods are DBSCAN, AGGLOMERATIVE, and KMEANS.

All the predefined spatial functions support computing the spatial weights and storing them in a datastore for later use. The goal of these functions is to execute common tasks involving spatial information. You can always add more functions for specific purposes as described in Store a Function for Embedded Execution.