What’s Artful and Scientific About Data?

Posted by Taylor Haney on Tue, May 19, 2015

describe the image
Excited to share some Data & Analytics themed blog posts with you in the coming weeks. Our 2015 MITX Data Summit Advisory board has contributed some fantastic posts to give you a taste of the exciting content that will be discussed at this year's Summit on June 23rd. Interested in attending? Register here. This guest blog post is written by Judah Philips, Author, "Building a Digital Analytics Organization".

describe the imageJudah Phillips, CEO and Founder of SmartCurrent, helps people improve their analytics in order to identify and take advantage of the business opportunities that significantly impact results, achieve goals, and create value. He helps companies find innovative ways to use data and analytics to improve efficiency, increase revenue, reduce cost, and boost profitability. Judah specializes in strategic reviews and the resulting work required for enhancing the effectiveness of your business. Phillips has built and managed business and digital analytics teams for global companies worldwide during the last 17 years.

In June you can join the MITX community at our 2015 Data Summit to discuss and learn about the “art and science of data.” But what does that really mean: “art and science?” I hear it all the time. I’ve even used the phrase. While I’ve always liked that phrase for describing analytics, it’s a phrase worth elaborating. So in this blog post, I’m going to unpack this oft-heard phrase and give you my interpretation related to some of the artful and scientific elements of data and analytics.

First, let’s define art. According to Google’s “define:” operation, it’s the “expression and application of human creative skill and imagination typically in visual form appreciated primarily for beauty and emotional power.” Fair enough. I certainly have seen the emotional power of a declining revenue ratio or increasing churn metric presented in a beautiful visualization, but is that enough to make it art? You decide; but I do think data and analytics can be artful in the following areas:

  • Business questions and requirements are the specific and critical items that must be executed in order to deliver successfully an analytics project. The questions and requirements provide the frame for technical activities and allow for the data and analysis to be relevant to the people (or systems) that will consume it. Gathering, synthesizing, documenting questions and requirements is artful.

  • Analytical interpretation has rules (math being not the least of them), but the interpretation is a product of the analyst who produced it. While rules must be followed, the brushstrokes are owned by the analyst, and interpretation is inherently creative even within formal guidelines. That’s one of the reasons why you can always question the analysis but not necessarily the data.

  • Data visualization is the presentation of data in pictures and images that not only tell a story in clear and concise way, but also makes data and information beautiful. While there are known types of and formats for data visualization, the approach taken by an analyst can vary greatly – and in that customized approach to visualization, the creation is the art.

  • Socialization and communication is the process for articulating and disseminating analytical outputs in a way that meets business requirements and answers the business questions. Undoubtedly, since all analysis can be political in context, the deftness and skill in which and when the interpretation and analysis is told to business stakeholders is crucial for ensuring the results are heard, understood, and actioned on. In fact, the importance of appropriate socialization and communication of analytical results to the perception of success with analytics can’t be underestimated.

Now let’s define “science.” Google tells us it is the “intellectual and practical activity encompassing the systematic study of the structure and behavior of the physical and natural world through observation and experiment.” Data and analytics are indeed intellectual activities that require a systematic approach focused on observation and experimentation about behavior and structure. The data science in analytics can most easily be understood in these areas:

  • Data collection is the technical process of generating, gathering, processing, and measuring data collected in a systematic manner that enables the answering of business questions, the testing of hypotheses, and evaluation of data to inform technical and business processes. The methods for accomplishing it are scientific.

  • Data wrangling is another technical process that can overlap with and is related to data collection, but is different enough to stand on its own. As practitioners know, data can be dirty, and to clean it up you must munge around in it and figure out how to create the data and relationships necessary for your analytical plan. Data wrangling, thus, is a term used to refer to a loose collection of processes and techniques that transform dirty data into clean data fit for analysis. Wrangling can involve manually (or automatically) converting and mapping data from one format to another, most often across multiple sources. Some might say this is art, but it’s mostly science.

  • Data modeling is the understanding of required data needed to answer a business questions and the identification of the relationships among the data in order to ensure they are fit for analytical purposes. A model will incorporate requirements and business activities and identify concepts about the data like entities, attributes, relationships, and integrity rules in order to create a logical model that can, in turn, be physically implemented in a technical environment (with the requisite tables, columns, fields, keys, indices and so on). Science.

  • Applied analysis. I’m using this term as a catch-all for the scientifically-derived statistical, mathematical, engineering, machine learning, and even artificial intelligence applied to data in order to analyze automatically and create outputs that can drive business value and profitable outcomes.

Data and analytics are, of course, art and science in other ways too, and the concepts aren’t mutually exclusive. Instead the art and science can be applied in what I have written about as the “Analytics Value Chain” in my book “Building a Digital Analytics Organization.” The Analytics Value chain, by its explanatory nature, incorporates the artful and scientific activities as a set of phases required to create analytical outcomes that help people realize positive business outcomes. For example, it starts with business questions and requirements (art) to yield specifications that define data collection (science) and modeling and then moves on to governance and into reporting, analysis, optimization, prediction and automation. Intrigued? Disagree? Fascinated? Agree? Then join us for these types of discussions and more at the 2015 MITX Data Summit. See you there.

Topics: MITX, data