Sunday, 27 November 2016

Getting Started with Graphlab - A Python library for Machine Learning

Before Starting with Graphlab, We have to configure our system with some basic tools such as Python, Jupyter Notebook etc. You can find 'How-To' on this link -

What is GraphLab ??
GraphLab Create is a Python library, backed by a C++ engine, for quickly building large-scale, high-performance data products. Some key features of GraphLab Create are:
  • Analyze terabyte scale data at interactive speeds, on your desktop.
  • A Single platform for tabular data, graphs, text, and images.
  • State of the art machine learning algorithms including deep learning, boosted trees, and factorization machines.
  • Run the same code on your laptop or in a distributed system, using a Hadoop Yarn or EC2 cluster.
  • Focus on tasks or machine learning with the flexible API.
  • Visualize data for exploration and production monitoring.
After the installation of Graphlab library we can use it as any python library.

Use Jupyter Notebook for starter, Open a Python notebook in Jupyter Notebook and execute below commands to see graphlab working -

 a. Importing Graphlab - 


b. Reading CSV file
This method will parse the input file and convert it into a SFrame variable


c. Getting Started with SFrame 

i. View content of SFrame variable sf


ii. View Head lines (top lines) 


ii. View Tail lines (last lines)

Like the below page to get update