Python Configuration

Gathr supports Python 3. To support Python 3, configure as mentioned below:

Python 3 (preferred 3.8.8) should be installed on all nodes of the cluster.

Python 3 should be the default python version on all the nodes.

Install Python 3 with the root user and create a soft link of the Python 3 binaries in /usr/bin, so that it will be available for all users.

Python 3 should be accessible with command, ‘python3.

All the required libraries should be installed for both python versions (not mandatory).

Python Libraries for Registering Models #

To register H2O, Scikit, or Spark ML models, and use them in scoring pipelines, follow the below steps:

For Python 3

1. Install python 3.8.8

2. Install pip3

3. Install H2O using the below command:

   pip3 install h2o==3.28.1.3
   

4. Install the scikit dependencies using the below command:

   root> pip3 install scikit-learn==0.24.1
   root> pip3 install numpy==1.19.2
   root> pip3 install pandas==1.2.4
   root> pip3 install matplotlib==3.4.2
   root> pip3 install mlflow==1.0.0
   root> pip3 install hdfs==2.5.8
   root> pip3 install scipy==1.6.2