Executing Python Pickle

What will you learn

1. How to use python pickle object inside NetFlow node/process.

2. How to apply pickle transformation on a process table.

After reading this tutorial (and its prerequisite) you should be able to include your custom python code that preforms an operation on a process table (transformation, predictions, …) and store resulting table back into process where it can further be used.

Prerequisites

• Access to NetiLab application with enough permissions.

• Python Server configured for user’s workspace.

• Read tutorial Uploading Custom Python Pickle/Model .

Introduction

NetFlow has various different nodes that applies a transformation on a process table. But in some use cases you would need some new transformation, or you have trained a ML model and would like to use it inside NetFlow processes. This can be done using Python and pickling operation, together with the ability to upload and use such objects inside NetFlow.

We will show this functionality with the pickle object created in Uploading Custom Python Pickle/Model .

Steps

Step 1

Open navigation menu inside NetFlow and click on New Process and select Blank process. Add node SQL Query (Single) and set Output Table setting to table. In the SQL Query field you can add the following

select 'test1' as Col1
union all
select 'test2' as Col1

which will create a table with two rows and one column.

Now save the process.

Step 2

In the same process and new node Python Execute Pickle Table and set the following configuration:

• Input Table: table

• Pickle: dummy_model_pickle

• Function: dummy_predict

• Output Table: predicted

Save the process again.

Step 3

Run the process. It should be successfully executed and at the end, there should be process table predicted with two column, where second column was generated with the pickle.