Skip links

Machine Learning: Data-driven insights fast using Zepl with Snowflake

Ryan Rechkemmer – Senior Analytics Data Engineer

As data and its applications continue to multiply – and with colleagues working remotely more and more – cloud-based data management and analysis have become increasingly valuable in solving organizations’ time-sensitive business problems.

Accordingly, the Right Triangle team has aligned with industry experts and providers to simplify data management and integrate it more easily into standard processes.

Many of today’s machine-learning tools contain roadblocks, including:

  • Slow setup
  • Intimidating interfaces
  • Configuring and standardizing software for various local environments
  • Awkward sharing of work or results

Conversely, Zepl is an entirely cloud-based tool providing a quick and easy remedy without the hassles of traditional local-install issues. You can code within minutes of signing up.

While this essay focuses on Python for machine learning, Zepl also has interpreters installed for Apache Spark, Angular and other languages. Zepl’s Python interpreter has commonly used machine learning libraries and packages such as NumPy, Pandas, Sklearn and TensorFlow pre-installed. If other libraries or packages are desired, they can be installed using Pip or Conda, just like in a standard Python installation.

After importing libraries with a few simple lines of code, Zepl can facilitate:

Importing the Data

Zepl has readymade connections for many different data sources, including a very intuitive one for Snowflake. Initiating a connection only requires entering the information for the Snowflake database. To pull the data into Zepl, simply write a SQL SELECT query.

Processing and Analyzing the Data

Within Zepl, data can be manipulated like any other Python installation by using NumPy and Pandas. Additionally, Zepl provides nice charting and graphics capabilities for displaying data and also supports many Python libraries for graphical displays.

Easily Dividing the Data into Training and Testing Subsets

Machine learning packages such as Sklearn require only a few lines of code to randomly separate data for training and testing.

Selecting and Testing the Model

Python machine learning packages go above and beyond here. An advantage to Zepl’s notebook format is that displaying results after each paragraph is more intuitive than standard development platforms. With the output so quickly accessible, tweaking and debugging are easier.

Insight Generation

The model is immediately useful. Dashboards can be quickly created and customized for specific audiences. Dashboards and charts with unique web addresses can be easily shared and are viewable by anyone with web browser access.

Although setting up a machine learning project in a new environment is usually cause for concern, Zepl provides a turnkey solution. And with the ability to code right away, results can be quickly achieved and shared.

Zepl is but one machine learning tool used by Right Triangle to help clients manage project timelines, large volumes of data and collaboration among colleagues. Paired with nearly infinite cloud storage, Zepl makes data accessible and actionable to help organizations achieve streamlined business solutions.