Issue link: https://industrialengineer.epubxp.com/i/777540
26 ISE Magazine | www.iise.org/ISEmagazine T systems engineering When it's big data and when it's not By Ricardo Valerdi The ongoing analytics revolution has come with a variety of buzzwords like data-driven, smart data and data science. Arguably the most prevalent buzzword is big data. Its broad use has also resulted in frequent misuse. Part of the challenge is that there is no official definition for big data; therefore, I will attempt to provide some boundar- ies that can help distinguish when it is appropriate to use the term. Naturally, there are gray areas that require hu- man judgment to determine when big data applies. Experts suggest that big data in- volves situations that are character- ized by the four V's: • Volume: Large data sets • Velocity: Data that need to be processed and acted upon quickly • Variety: Data that are in different formats or come from different sources • Veracity: Data that are uncertain The four V's are relative, of course. What might be large or fast in one in- dustry might be small and slow in an- other. More important are the tools that need to be brought to bear in order to manage and benefit from the availability of data. One of the causes of big data is the emergence of the digital age. Every second there are an unknown number of devices and sensors generating data. Data are being accumulated at a faster rate than ever before thanks to the bil- lions of smartphones, wearable devices, web pages, apps, stock fluctuations, news articles, instant messages, social media posts, videos and photos con- nected to the cloud (e.g., the internet of things). For example, every 60 seconds there are 98,000 tweets, 695,000 Facebook status updates, 11 million instant mes- sages, 698,445 Google searches, 168 million emails sent and 217 new mobile web users. These examples unmistak- ably fit into the big data definition. Other situations may sound like big data, but they actually are not. For instance, if a company is working to obtain insights from its customer base by mining its Twitter feeds, its dataset might not be very large compared to the total number of daily tweets (vol- ume), the analysis would take hours or days to perform (velocity), the infor- mation would be text-based (variety), and the data would be fairly reliable since it comes directly from the source (veracity). Similarly, we can consider one of the largest repositories of information on the internet: Wikipedia. All of its text data could fit on a single USB drive (volume), does not change often (velocity), is con- strained to text and photos (variety) and is frequently peer-reviewed (veracity). Regardless of how well a situation fits the definition big data, what is impor- tant is how the analyses applied can help make decisions. This is where industrial and systems engineers come in. Finding patterns in the data, separat- ing the signal from the noise, us- ing data to predict future outcomes and dealing with unstructured data are some of the ways in which we can provide value. ISE programs around the world have been teach- ing these skills for decades. To many practicing engineers and statisticians, these are new names to well-known concepts. Anyone who understands what to do with data and how to apply statistical tools knows how these concepts can support decision-making. As more devices come online and our environment becomes even more con- nected, knowing what qualifies as big data can help us determine the necessary tools and techniques that should be ap- plied to make sense of it all. Ricardo Valerdi is a ssociate professor at the U iversity of Arizo a i the Departme t of Systems a d I dustrial E gi eeri g. He is the former co-editor-i -chief of the Journal of Enterprise Transformation, a collabora- tive ve ture betwee ISE a d INCOSE. Reach him at rvalerdi@arizo a.edu. To many practicing engineers and statisticians, these are new names for well-known concepts.