Tools and techniques for management reporting and analysis have evolved since computers first came out. One can argue that the first management reporting tool was COBOL (COmmon Business Oriented Language). It allowed business people to get data out of systems created by computer people. COBOL was eventually replaced for reporting by tools like Information Builders (IBI), which started as programming languages used for reporting.
Spreadsheets appeared as a way to analyze and present data. Originally they couldn’t access databases but that has long since changed.
Since then we moved onto decision support systems and business intelligence, with tools like Business Objects, MicroStrategy, Tableau, Qlik, Cognos, and a ton of others.
What almost all of these tools have in common is that they let non-technical managers access and and present data. For example, the sales manager using Tableau can grab sales data for the past two years, see it presented in a pivot table showing the top customers by region and in a chart showing the trend in overall sales. However, almost all applications of these business intelligence tools are backward looking. They present summaries of the data we own.
The next step in this reporting evolution is data science. Data science is more about identifying correlations, calculating probabilities, and predicting the future than reporting the past. Thus, one may use a data science model to predict the likelihood that a customer will purchase based on their zip code (and, in most cases, a host of other factors).
While it’s not initially obvious, prediction extends into realms like facial recognition. Most facial recognition is performed using data science tools and data about people the system already knows, data like photographs tagged with people’s names. These systems don’t know that you are who you say you are but they calculate a likelihood that you are that person (to reinforce the point that this is simply a prediction, and that predictions sometimes fail, here’s an article about a study finding that an Amazon facial recognition tool misidentified 28 members of Congress).
So, the major difference between data science and business intelligence is this focus on being forward, rather than backward, looking. However, there are a few other differences.
TOOL SETS: As you might expect, data scientists use different tools than do BI users. Whereas most BI tools provide users with point and click interfaces to access and format data, data science tools tend to be programming languages, like Python or R, with add-in libraries of code tailored to specific problems.
AUDIENCE: While vendors are developing end-user-accessible data science tools, the vast majority of data science technologies require strong technical skills as well as a background in statistics, math, or a related field. As a result, data science technologies are used by a very limited set of people. These people are called upon to build ‘predictive models’, the results of which are distributed to much broader audiences.
PROGRAMMING COMPETENCE: Because BI tools are intended to give easy data access to managers, they do not require much technical skill (beyond that needed to initially configure them). Data science tools, however, do require programming chops. There are differences between the job of a programmer and the job of a data scientists but that’s a topic for another post.
So, if you’re just getting into this field, what do you need, business intelligence or data science? This is obviously not a simple question to answer. You likely want to start with business intelligence as almost all companies moving into data science have. Not only will it give your managers access to their data but BI tools will be used by your future data scientists as they investigate that data. Once you’re comfortable with these tools, then take the leap into data science. Remember, whatever you do, purchasing technologies just to ‘keep up with the Joneses’ is ill-advised. Start with a clear picture of what you’re expecting and know what it is you want to do that you can’t do today.
Finally, if you have any questions, reach out to Dataspace! We’d love to help you navigate this powerful yet complex space!