Create and Store a User-Defined Python Function

10.4.7.2 Create and Store a User-Defined Python Function

Use the oml.script.create function to add a user-defined Python function to the script repository.

With the oml.script.create function, you can store a single user-defined Python function in the OML4Py script repository. You can then specify the user-defined Python function as the func argument to the Embedded Python Execution functions oml.do_eval, oml.group_apply, oml.index_apply, oml.row_apply, and oml.table_apply.

You can make the user-defined Python function either private or global. A private user-defined Python function is available only to the owner, unless the owner grants the read privilege to other users. A global user-defined Python function is available to any user.

The syntax of oml.script.create is the following:

oml.script.create(name, func, is_global=False, overwrite=False)

The name argument is a string that specifies a name for the user-defined Python function in the Python script repository.

The func argument is the Python function to run. The argument can be a Python function or a string that contains the definition of a Python function. You must specify a string in an interactive session if readline cannot get the command history.

The is_global argument is a boolean that specifies whether to create a global user-defined Python function. The default value is False, which indicates that the user-defined Python function is a private function available only to the current session user. When is_global is True, it specifies that the function is global and every user has the read privilege and the execute privilege to it.

The overwrite argument is a boolean that specifies whether to overwrite the user-defined Python function if it already exists. The default value is False.

Example 10-11 Using the oml.script.create Function

This example stores two user-defined Python functions in the script repository. It then lists the contents of the script repository using different arguments to the oml.script.dir function.

Load the iris dataset as a pandas dataframe from the seaborn library. Use the oml.create function to create the IRIS database table and the proxy object for the table.

%python

from sklearn import datasets
import pandas as pd
import oml

# Load the iris data set and create a pandas.DataFrame for it.
iris = datasets.load_iris()

# Create objects containing data for the user-defined functions to use.
x = pd.DataFrame(iris.data,
                 columns = ['Sepal_Length','Sepal_Width','Petal_Length','Petal_Width'])
y = pd.DataFrame(list(map(lambda x:
                           {0: 'setosa', 1: 'versicolor', 2:'virginica'}[x], iris.target)),
                 columns = ['Species'])
                 
# Create the IRIS database table and the proxy object for the table.

try:
  oml.drop(table="IRIS")
except:
  pass
  
oml_iris = oml.create(pd.concat([x, y], axis=1), table = 'IRIS')

Create an user-defined function build_lm1 and use oml.script.create function to store it in the OML4Py script repository. The parameter "build_lm1" is a string that specifies the name of the user-defined function. The parameter func=build_lm1 is the Python function to run. Run the user-defined Python function in embedded Python execution.

%python

# Define a function.

build_lm1 = '''def build_lm1(dat):
  from sklearn import linear_model
  regr = linear_model.LinearRegression()
  import pandas as pd
  dat = pd.get_dummies(dat, drop_first=True)
  X = dat[["Sepal_Width", "Petal_Length", "Petal_Width", "Species_versicolor", "Species_virginica"]]
  y = dat[["Sepal_Length"]]
  regr.fit(X, y)
  return regr'''

# Create a private user-defined Python function.
oml.script.create("build_lm1", func=build_lm1, overwrite=True)

# Run the user-defined Python function in embedded Python execution
res = oml.table_apply(oml_iris, func="build_lm1", oml_input_type="pandas.DataFrame")

res
res.coef_

The output is the following:

array([[ 0.49588894,  0.82924391, -0.31515517, -0.72356196, -1.02349781]])

Define another user-defined function build_lm2, store the function as a global script in the OML4Py script repository. Run the user-defined Python function in embedded Python execution.

%python

# Define another function

build_lm2 = '''def build_lm2(dat):
  from sklearn import linear_model
  regr = linear_model.LinearRegression()
  X = dat[["Petal_Width"]]
  y = dat[["Petal_Length"]]
  regr.fit(X, y)
  return regr'''

# Save the function as a global script to the script repository, overwriting any existing function with the same name.
oml.script.create("build_lm2", func=build_lm2, is_global=True,  overwrite=True)

res = oml.table_apply(oml_iris, func="build_lm2", oml_input_type="pandas.DataFrame")
res

The output is the following:

LinearRegression()

List the user-defined Python functions in the script repository available to the current user only.

%python

oml.script.dir()

The output is similar to the following:


                name  ...                date
0          build_lm1  ... 2022-12-15 19:02:44
1          build_mod  ... 2022-12-12 23:02:31
2      myFitMultiple  ... 2022-12-14 22:30:43
3  sample_iris_table  ... 2022-12-14 22:21:24

[4 rows x 4 columns]

List all of the user-defined Python functions available to the current user.

%python

oml.script.dir(sctype='all')

The output is similar to the following:


     owner  ...                date
0   PYQSYS  ... 2022-02-11 06:06:44
1   PYQSYS  ... 2022-10-19 16:59:50
2   PYQSYS  ... 2022-10-19 16:59:52
3   PYQSYS  ... 2022-10-19 16:59:53

List the user-defined Python functions available to all users.

%python

oml.script.dir(sctype='global')

The output is similar to the following:


                   name  ...                date
0                 GLBLM  ... 2022-02-11 06:06:44
1         RandomRedDots  ... 2022-10-19 16:59:50
2        RandomRedDots2  ... 2022-10-19 16:59:52
3        RandomRedDots3  ... 2022-10-19 16:59:53
4                  TEST  ... 2021-08-13 17:37:02
5                 TEST4  ... 2021-08-13 17:42:49
6              TEST_FUN  ... 2021-08-13 22:38:54