Tech Talk

Recent posts

Enterprise System Pitfalls: Summary
Today I’m wrapping up a series of posts on the broad topic of Enterprise System Pitfalls. In this series, my hope was to help shed light on the primary problems that cause us to miss budgets, fall short on capabilities, or completely fail when implementing an enterprise system. 

The Year in Review
 
As 2019 comes to a close, it’s time to count our blessings. One of mine has been the privilege (and fun!) of being able to reach out to so many interesting companies and get them to tell me what they’re doing that’s different, disruptive, and game-changing. The list of things I have to write about in future columns has only gotten longer in the nine months since I started writing this column.

Sustainable Innovation
 
Sustainability can yield multiple benefits to hotels. Saving energy and water yields direct cost savings. Revenue can be generated by guests who prefer to deal with businesses that minimize their environmental impact. And many would argue that conserving scarce resources is simply the right thing to do.

Meetings Innovation
 
The sale and delivery of groups and meetings is perhaps the most significant and under-automated functions for many hotels. Even though groups often account for 30% to 60% of revenue, most group bookings are still handled manually for most if not all of steps, as they move from a meeting planner’s research to a confirmed booking.

The biggest enemy to any system is complexity. In a system of inputs and outputs, such as an enterprise system, more complexity means more parts are used in interaction with inputs to create the outputs. Every part that must be built and maintained costs time and money



want to read more articles like this?

want to read more articles like this?

Sign up to receive our twice-a-month Watercooler and Siegel Sez Newsletters and never miss another article or news story.

x
 

Big Data: Correlation Does Not Equal Causation

08/28/2014
by Amy Maloney
 
There are many opinions on what constitutes big data, but the most widely accepted definition was coined by Doug Laney (currently with Gartner) back in 2001.  His definition of big data is based on the three V’s– volume, velocity and variety.
 
Since the introduction of the three V’s, industry experts have added at least two more V’s to the definition– value and variability– and more will likely follow.

From Laney’s definition, one might think that big data is relatively new.  According to Gil Press in his article A Very Short History of Big Data, the first attempts at quantifying volumes of data happened 70 years ago. 

Regardless of the definition, big data is here to stay.
     
For the hospitality industry, big data can take many forms and new data sources are becoming available all the time.  While new data sources should be explored, it is also important to make sure you are getting the most value possible from the data sources you already have.

While big data, or any data for that matter, can provide extremely useful information, the challenge lies in identifying the real business value it will provide and then making sure it is actionable.  This can be costly and involves both domain insight and advanced data analytics.  Given that big data is good at detecting correlation, it may often yield correlations that do not equal causation.

For example, as Sendhil Mullainathan (professor of Economics at Harvard University) writes in Hold the Phone: A Big-Data Conundrum, “sometimes data reveals only correlations and not conclusions.”  After complaining to his students about how his iPhone seemed to always slow down just before a new iPhone was released, one student decided to find out if there was any basis in his complaint. 

The research included cross-referencing searches using Google Trends for both “iPhone slow” and “Android slow” with release dates of both devices.  Looking at the data (screenshots below from Mullainathan’s article), it is tempting to conclude that Apple is somehow causing the phones to slow down when they release a new device as searches for “iPhone slow” spike at release dates while searches for “Android slow” are much more consistent. Further, given that Apple sells the device and controls the OS while Google only controls the OS and doesn’t make money directly from the hardware, it makes that assumption even more compelling.  However, when you factor in the percentage of iPhone users that religiously upgrade their OS (90 percent) to the percentage of Android users who do (18 percent), the assumption is no longer so convincing.  The slowness in old iPhones when new models are released could instead be a side effect of the new OS having been optimized for the new device.
         
Google Flu Trends, which aggregates Google searches in an attempt to accurately predict flu activity, experienced a similar conundrum.  In a Science magazine article published in March of 2014, researchers from Northeastern University and Harvard reported that Google’s flu-tracking service has consistently overestimated the number of flu cases in the United States. In addition, the algorithm completely missed the swine flu epidemic.

Although the researchers assert that “there are enormous scientific possibilities in big data,” they called Google Flu Trends an example of “big data hubris,” that is described as “the often implicit assumption that big data are a substitute for, rather than a supplement to, traditional data collection and analysis.”  

But, this also doesn’t mean that it is entirely coincidental or that correlation doesn’t at least suggest causation. A strong correlation gives us a reason to dig deeper, albeit with a healthy dose of skepticism. Why, because there is a natural human tendency to give in to confirmation bias (a type of selective thinking whereby one tends to notice and look for information that confirms one’s own beliefs).  Avoiding this bias requires not only critical thinking, but also actively seeking both supporting and contradictory evidence. Critical thinking and the application of sound scientific methods is something our team of pricing and data scientists take very seriously before formalizing new data sources into our algorithms. 

As the hospitality distribution landscape becomes increasingly complex to navigate, it is becoming ever more challenging for hoteliers to grow profits. While new data sources may be essential to future success and lead to new insights that will help drive profitability, it is important to learn from examples like Google Flu Trends and the iPhone theory. Applying new data sources to pricing and marketing strategies without including more conventional data analysis might have unintended consequences. Just because it is new and big, doesn’t automatically make it useful. Nor should it overshadow existing, proven methods for determining pricing and marketing strategies.

About The Author
Amy Maloney
VP of Hotel and Gaming Solutions
The Rainmaker Group


As the product manager and technical implementation lead for The Rainmaker Group’s gaming and hospitality business unit, Amy is responsible for defining the product strategy, managing vendor relationships, delivering innovative solutions and ensuring customers are receiving the most value from our products and services. Prior to joining Rainmaker, Amy led a team of engineers responsible for delivering and supporting intranet services and applications for NCR's retail solutions division and fulfillment operations.

 
Comments
Blog post currently doesn't have any comments.
Leave comment



 Security code