WINTER WARRIOR CHALLENGE

AccuWeather, the global leader in weather information and digital media, and Microsoft Corp. continue their strong collaboration with the release of the Winter Warrior Challenge, WinterWarrior.AccuWeather.com, an online challenge between two individuals to prove who had the worst winter and recognize the true Winter Warrior. Through this engaging new online tool, users will accurately and scientifically know if their childhood winters were snowier than friends or family as they compare winter weather impacts through an interactive display. The true purpose of the Winter Warrior Challenge is a fun, interactive way to show an application of our weather data coupled with analytics to help clients run their businesses more efficiently to save time and money.

Winter Warrior Challenge started from the humble beginnings of a team meeting on November 15, 2016 with a brainstorming session and me being charged with coming up with more ideas to present as "interesting visuals." With winter fast approaching, we came up with various ideas to discuss. These ranged from "my grandfather said that he walked to school up hill, both ways, in the snow" to looking at chances for a white Christmas or snow day, to giving people the most extreme weather for today for their locations. We settled on the former, and Winter Warrior was born!

By mid December 2016, I had completed my mock up of the Winter Warrior Challenge in Microsoft Power BI. We shared this with Microsoft, and a third party who would build the website. Everyone loved the concept and the prototype. From here, it was a collaborative effort between AccuWeather Analytics, AccuWeather Marketing, Microsoft, and the third party. One of the most gratifying moments for me came from learning that my visual prototype of the website saved the project 30% of cost and many hours of work. This was vastly important in trying to get Winter Warrior launched during the winter.

Due to a few complications, the Winter Warrior Challenge did not launch in early February as had been planned. It went live on March 6, 2017 just days before a massive snow storm pummeled the East Coast of the United States. As of March 29th, 2017, Winter Warrior has experienced 47k hits 23 days since launch. People are enjoying the site, and hopefully it will see a resurgence in popularity when we as a society start thinking about winter again later this year, and AccuWeather re-launches it at the start of Winter 2017. 

Below you will find the process of how we went from nothing to the Winter Warrior Challenge! [Click on images to enlarge]

brainstorming

During a team meeting, I was tasked with brainstorming ideas for an interesting visual that would showcase AccuWeather, our data, and analytics capabilities. I researched what was out there that was weather related, especially any good "this day in history" stuff. Nothing great was out there but without becoming a far more extensive project, this was too broad for what my bosses were looking for. As a team, we started discussing winter related ideas for "Project: Interesting Visuals." Ideas included: Chance of a white Christmas? When is your best chance to build a snowman in the coming days/weeks? Chances of snow days for kids. What's the best day to go outside and enjoy fresh air during the winter months? And who had the worst winter as a child? 

We thought the Winter Warrior Challenge was the most compelling of our ideas, and I was off to explore the data to build this concept and prototype how it would look and feel. 

Data Exploration

One of the first steps I encountered was looking at the location data for our historical snow data. We would be using METAR station data for the snowfall amounts, and I needed to look at the distribution of METAR stations to cities both in the United States, and globally. I used Esri's ArcMap to map the weather stations and cities including visually displaying city population size to determine more important location coverage.

plotting/joining cities

My first step was finding the needed global and US cities' shapefiles, and plotting in ArcMap. I had a global cities map with mostly major cities and a separate data set of all US cities. To begin, I applied a visual symbology to the global cities so I would know what the most important were. I removed the US cities from the global cities' dataset. Then I selected out US cities under 5000 citizens. I chose to use US cities with populations of 5000+ due to coverage issues especially in the Great Plains and Rockies. Finally, I joined the two separate cities datasets into one.

joining cities with metar stations

My next major step was using a spatial join to join the cities to the nearest METAR station. This spatial join was based on the nearest weather station to the cities' latitude/longitude. The METAR stations would provide the historical snow data needed for calculating someone's Winter Warrior score.

Transforming image to shapefile

Finally, I found an image for possible snowfall for the globe. This made it easier to QC our historical snow data plus remove all major Global Cities that will never have snowfall and code them as a 0 on the Winter Warrior scale. To be able to select out cities not falling in potential snow regions, the top image had to be georeferenced, rasterized, and finally made into a polygon shapefile. This method still required some manual data manipulation of the resultant polygons to make the shapefile useable, but this was far more accurate than hand drawing a snow cover polygon. 

Power BI Prototype

After exploring the data, making decisions regarding cities, and joining the selected cities to the closest METAR/UFDB weather stations, it was time to mock up how we wanted the "website/app" to look and feel like for the end user. It was important to my team and I to showcase how we wanted the end product to feel and the emotions it would evoke. I used Microsoft Power BI to accomplish this so that I could further my skills with the software, and because we were exploring how to use Power BI for this and future AccuWeather Powered by Microsoft products. We also made the decision to base the Winter Warrior score on an individual's winter snow data from the age of 8 through 11. This would provide four years of data for averaging their overall Winter Warrior score. The snow amount totals were then ran through a disruption algorithm to determine each individual's Winter Warrior score. Someone growing up in Dallas, receiving 6 inches of snow is a much bigger deal than someone growing up in Denver receiving 6 inches of snow. This piece of the puzzle was important because we wanted to show beyond just basic weather metrics but a use case for analytics. 

Landing page

Who is the Winter Warrior? Includes the old saying we have all heard from our parents and grandparents.

add information

This is where to put in your information and a second person's information. Also gives an explanation of how your Winter Warrior score will be determined.

results

The results page will include not only a graph showing each person's Winter Warrior score, but also explanation of the score not being just basic snow data. Also includes social media links.

WINTER WARRIOR CHALLENGE GOES LIVE!

AccuWeather management and Microsoft loved the Winter Warrior Challenge. They tapped a third-party company to build the website. And from here, it was a collaborative effort between AccuWeather Analytics, AccuWeather Marketing, Microsoft, and the third-party website developer. One of the most gratifying moments for me came from learning that my visual prototype of the website saved the project 30% of cost and many hours of work. Below are images of the live site on a mobile device. I hope this accurately portrays how this concept went from an idea on paper to a live, interactive website, and my role in its development. Check out the Winter Warrior Challenge for yourself!