January 21

Machine Learning Applications Utilizing Weather Data

Artificial intelligence (AI) and its subsets, machine learning (ML)  and deep learning (DL) are grabbing all the headlines these days.  From basic job replacement to a possible robot uprising, reporters, influencers & futurists are waxing poetically about the benefits (and threats) future generations will experience due to advances in these technologies.

 

Diagram showing layers of machine learningAI/ML/DL Hierarchies

While the Team at AerisWeather follows those storylines along with everyone else in the technology world, we’re also fortunate enough to have a front row seat to some of the CURRENT applications of these exciting toolsets.  Today, we work with a broad array of engineers, software developers and data scientists across industries as they put AI/ML/DL technology to work solving today’s real business challenges.

How did we land this ticket in the front row?  The answer lies in the data – environmental data to be precise.  While there are plenty of use cases where weather & other environmental factors do not affect the outcomes of machine-powered technology (think manufacturing robot), the reality is that the majority of use cases have an environmental component that may affect the proper training of an algorithm (e.g. the temperature affected the performance of equipment) or, the ability to execute on a task (e.g. the road is icy ahead, slow the car).

Here’s a quick spin through various industries and some compelling use cases where factors such as temperature, precipitation, wind and lightning were key contextual inputs to the tasks at hand.

UTILITIES:  This is fun place to start – with an industry that’s known for being conservative and not perceived as a leader in the next technology revolution.  Don’t sell this crowd short though because they’re getting it done in some interesting areas.

  • Solar Power Use Case: Historically, proximity to customers, availability of land, and days of sunshine have driven the site selection & viability of a utility-scale solar field. Today, energy engineers use those legacy practices along with machine learning algorithms that leverage solar radiation (available even on cloudy days) and ambient conditions (wind, humidity, temp) to optimize & accurately forecast the performance of their installations.
  • Transmission Cable Failure Use Case: The fires in California have made it painfully clear that adverse environmental conditions can accelerate the degradation of equipment.  Utilities are now looking backwards to model how environmental events such as fire, high heat and lightning led to the failure of assets in the field.  With weather forecasts and advanced machine learning models, utilities are now better positioned to predict the failure of assets and proactively replace critical infrastructure before disaster strikes.

MARKETING: A cutting-edge, data-driven industry where customer experience, content, and ads are automated to optimize each person’s unique digital world. If there’s an industry willing to flesh out a new idea this is it.

  • Smart Ad Placement Use Case: It’s no surprise that when temperatures dip below historical norms our minds wander to far away beaches and palm trees. Marketing professionals know this and have proven successful by increasing ad spend during those forecasted periods and altering content to emphasize specific destinations for travel and hospitality clients.
  • Regression Analysis Use Case: Retail marketing and business intelligence professionals bring all the facts to the table by conducting deep learning to uncover business phenomena against first party data. By integrating contextual historical weather data with a data lake or warehouse, these retailers have uncovered trends not just in specific product sales but in operational efficiencies and the staffing requirements to meet high customer demand.  Now, retailers are not just placing ads tied to weather forecasts, they’re adjusting inventory and store headcount in advance of known opportunities.

SMART HOME: Behind all those cool displays, gadgets & switches working their way into your home are some super smart engineers and data scientists.  When they add hyper-local contextual data like weather to the equation, the result is a user-experience that feels like the set of a sci-fi movie.

  • Indoor Air Quality Use Case:  With new homes being built “tight” to prevent outdoor air from seeping in, the unintended consequences include holding air in that can often be bad for our health.  Homeowners have options today however thanks to machine learning technology that leverages indoor IoT sensors and external air quality data.  These highly accurate algorithms are now monitoring real-time data streams in homes, highlighting when indoor air quality is harmful and automating the filtration systems that can keep us safe.
  • Irrigation Use Case:  You’re heard of a Smart Home but what about a Smart Sprinkler? That dream is a reality in backyards across the country thanks to IoT-enabled devices using hyper-local weather data to control legacy irrigation systems.  These smart systems now incorporate the past, present and future to determine whether to open the valve on a yard optimized for water conservation.

TRANSPORTATION AND LOGISTICS: If there’s one industry that is particularly exposed to adverse weather conditions, it’s the professionals responsible for moving goods via highways, maritime or air.  Not only do these transportation companies need to strive for on-time delivery and safe working conditions for their employees, they often need to insulate their goods from extreme weather conditions such as heat, cold and humidity.

  • Packaging Logic Use Case: If there’s something that most of us take for granted as place online orders – whether it be books, bottled water or blueberries – is that it’s all delivered via the same truck and with no issue. How is this? Perishable good companies for example, are leveraging a blend of historical weather data with past shipping reports to train algorithms that inform their packaging needs – ice, insulation, transit speed – ensuring you are never disappointed.
  • Routing Optimization Use Case:  In the past, the responsibility for minimizing route times while maintaining high safety standards often fell to the driver – where experience was often the key to safe, on-time arrivals.   Unfortunately, that model is not scalable.  Enter today’s technology tools, where routes are optimized in real time in the cloud, often around inclement weather and poor road conditions.  With the aid of current conditions, forecasts, alerts and radar layers in their routing software, transportation companies are delivering more goods, faster and safer than ever before.

Why AerisWeather?

With a team of dedicated weather data experts to support you in your artificial and machine learning endeavors, AerisWeather has both the experience and resources to help you succeed. Whether it’s a small, experimental application or, an advanced, enterprise-scale solution, we have the global historical, current and forecasted data to power your AI/ML/DL algorithms. Please contact our team or start with a free developer account today.

Share this post:

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.