The data scientist position may well be one of the most multi-faceted jobs around, involving aspects of statistics, practical business knowledge, interpersonal skills, programming languages, and many other wide-sweeping qualifications.
For business end users of data-driven processes, however, many times these professionals simply seem like glorified IT personnel: theyâre the new hirees the former goes to, and waits upon, to get the data required to do their jobs.
Today, analytics platforms featuring conversational, interactive responses to questions can eliminate the backlog of demands for data while transforming the business into citizen data scientists capable of performing enough lower echelon data science functions to conduct their own analytics.
Moreover, they also equip them with the means to modify the data and answer their own questions, as needed, to take a greater sense of ownership and perhaps even pride in the data which impacts their jobs.
Ben Szekely, Cambridge Semantics Vice President of Solutions and Pre-Sales, reflected that, âBecause business users are getting answers back in real time theyâre able to start making judgments about the data, and theyâre developing a level of trust and intuition about the data in their organization that wasnât there before.â
Real-Time Answers, Ad-Hoc Questions
The most immediately demonstrable facet of a citizen data scientist is the ability to answer oneâs own data-centric questions autonomously. Dependence on external IT personnel or data scientists is not required with centralized data lake options enhanced by smart data technologies and modern query mechanisms. This combination, which leverages in-memory computing, parallel processing, and the power of the cloud to scale on demand, exploits the high-resolution copy of data assets linked together within a semantic data lake. Users are able to issue their own questions and answers of the resulting enterprise knowledge graph through either a simple web browser interface or their favorite self-service BI tool of choiceâthe latter of which is likely already in use at their organization. âTheyâre getting their answers through a real-time conversation and interaction with the content, versus going and asking someone and getting back a Powerpoint deck,â Szekely mentioned. âThatâs a very static thing which they canât converse with or really understand necessarily, or [understand] how the answer was come to.â
Understanding Answers and Data
Full-fledged data scientists are able to trust in data and analytics results because they have an intimate knowledge of those data and the processes they underwent to supply answers to questions. Citizen data scientists can have that same understanding and readily determine insight into data provenance. The underlying graph mechanisms powering these options deliver full lineage of dataâs sources, transformation, and other aspects of their use so citizen data scientists can retrace the dataâs journey to their analytics results. Even laymen business users can understand how to traverse a knowledge graph for these purposes, because all of the data modeling is done in terms predicated on business concepts and processesâas opposed to arcane query languages or IT functions. âWe talk about the way things are related to basic concepts and properties,â Szekely said. âYou donât have to be able to read an ER diagram to understand the data. You just have to be able to look at basic names and relationships.â Those names and relationships are described in business terms to maximize end user understanding of data.
Selecting and Modeling Data
Another key function of the data science position is determining which sources are relevant for questions, and modeling their data in a way so that a particular application can extract value from them. Citizen data scientists can also perform this basic functionality autonomously with a number of automated data modeling features. Relational technologies, for example, require copious time periods for constructing data models, calibrating additional data to fit into predefined schemas, and successfully mapping it all together. They require data scientists or IT to âbuild that monolithic data warehouse model and then map everything in it,â Szekely acknowledged. Conversely, smart data lakes enable users to begin analyzing data as soon as they are ingested, without having to wait for data to be formatted to fit the schema requirements of the repository. There are even basic data cleaning and preparation formulas to facilitate this prerequisite for citizen data scientists. According to Szekely, âYou can bring in new data and weâll build a model kind of automatically from the source data. You can start exploring it and looking at it without doing any additional modeling. The modeling comes in when you want to start connecting it up to other sources or reshaping the data to help with particular business problems.â
Enterprise Accessible Data Science
Previously, data science was relegated to the domain of a select few users who functioned as gatekeepers for the rest of the enterprise. However, self-service analytics platforms are able to effectively democratize some of the rudimentary elements of this discipline so business users can begin accessing their own data. By turning business users into citizen data scientists, these technologies are helping to optimize manpower and productivity across the enterprise.