June 12th, 2018

Time is Money! Automate Your Time-Series Forecasts with Driverless AI

RSS icon RSS Category: Driverless AI
Details about H2o ai experiment

Time-series forecasting is one of the most common and important tasks in business analytics. There are many real-world applications like sales, weather, stock market, energy demand, just to name a few. We strongly believe that automation can help our users deliver business value in a timely manner. Therefore, once again we translated our Kaggle Grand Masters’ time-series recipes into our automatic machine learning platform Driverless AI (version 1.2). This blog post introduces the new time-series functionality with a simple sales forecasting example.
The key features/recipes that make automation possible are:

  • Automatic handling of time groups (e.g. different stores and departments)
  • Robust time-series validation
    • Accounts for gaps and forecast horizon
    • Uses past information only (i.e. no data leakage)
  • Time-series specific feature engineering recipes
    • Date features like day of week, day of month etc.
    • AutoRegressive features like optimal lag and lag-features interaction
    • Different types of exponentially weighted moving averages
    • Aggregation of past information (different time groups and time intervals)
    • Target transformations and differentiation
  • Integration with existing feature engineering functions (recipes and optimization)
  • Automatic pipelines generation (see this blog post)

A Typical Example: Sales Forecasting

Below is a typical example of sales forecasting based on Walmart competition on Kaggle. In order to frame it as a machine learning problem, we formulate the historical sales data and additional attributes as shown below:
Raw data:
Table for store sales
Data formulated for machine learning:
Table for departement_strore
Once you have your data prepared in tabular format (see raw data above), Driverless AI can formulate it for machine learning and sort out the rest. If this is your very first session, the Driverless AI assistant (new feature in version 1.2) will guide you through the journey.
Alert for driverless ai
Similar to previous Driverless AI examples, users need to select the dataset for training/test and define the target. For time-series, users need to define the time column (by choosing AUTO or selecting the date column manually). If weighted scoring is required (like the Walmart Kaggle competition), users can select the column with specific weights for different samples.
Details about H2o ai experiment
If users prefer to use automatic handling of time groups, they can leave the setting for time groups columns as AUTO.
simple settings
Expert users can define specific time groups and change other settings as shown below.
Data about simple settings
Once the experiment is finished, users can make new predictions and download the scoring pipeline just like any other Driverless AI experiments.
Walmart demo data
Seeing is believing. Try Driverless AI yourself today. Sign up here for a free 21-day trial license.
Until next time,
Joe
Bonus fact: The masterminds behind our time-series recipes are Marios Michailidis and Mathias Müller so internally we call this feature AutoM&M.

About the Author

Jo Fai Chow
Jo-Fai Chow

Jo-fai (or Joe) has multiple roles at H2O.ai. He is best known as the #360Selfie guy nowadays. On LinkedIn, he is the data science evangelist and community manager but everyone knows that his photography skills totally overshadow his data science knowledge these days. On Twitter, he sounds like a die-hard MATLAB fanboy with the handle @matlabulous (because MATLAB was his favourite tool at Uni). Since joining H2O.ai in 2016, Joe has delivered H2O talks/workshops in 40+ cities around Europe and US. Before joining H2O, he was in the business intelligence team at Virgin Media where he developed data products to enable quick and smart business decisions. He also worked remotely for Domino Data Lab as a data science evangelist promoting products via blogging and giving talks at external events.

Leave a Reply

Novel Ways Using Driverless AI

I am biased when I write that Driverless AI (DAI) is amazing, but what's more

November 14, 2019 - by Thomas Ott
Useful Machine Learning Sessions from the H2O World New York

Conferences not only help us learn new skills but also enable us to build brand

November 13, 2019 - by Parul Pandey
Fallback Featured Image
Accelerate Machine Learning workflows with H2O.ai Driverless AI on Red Hat OpenShift, Enterprise Kubernetes Platform

Organizations globally are operationalizing containers and Kubernetes to accelerate Machine Learning lifecycles as these technologies

November 12, 2019 - by Nicholas Png
Image Tasks on H2O Driverless AI

I’d like to thank Grandmaster Yauhen Babakhin for reviewing the drafts and the very useful

November 12, 2019 - by Sanyam Bhutani
Importing, Inspecting, and Scoring With MOJO Models Inside H2O

Machine-learning models created with H2O may be exported in two basic ways: Binary format, Model Object, Optimized

November 8, 2019 - by Pavel Pscheidl
Natural Language Processing in H2O’s Driverless AI

Note: I’d like to thank Grandmaster SRK for a lot of suggestions and corrections with

November 6, 2019 - by Sanyam Bhutani

Join the AI Revolution

Subscribe, read the documentation, download or contact us.

Subscribe to the Newsletter

Start Your 21-Day Free Trial Today

Get It Now
Desktop img