Many analytics groups are currently evaluating data science technologies for their enterprise analytic platform, and they quickly realize the number of technologies available is overwhelming. Selecting the right technology for your group requires more than evaluating the features and price of individual products. Selecting the right technology requires aligning both your current and future talent pool as well as your internal development and data engineering processes.
Engineering vs GUIs – What should we invest in?
I hate using the phrase ‘people, process, technology’, but there is a reason it is so popular. When evaluating data science technologies, or the need for data science technologies, consider the following:
Data science technologies, like Alteryx and Dataiku, market themselves as technologies that help organizations overcome the data science talent shortage. While I do not believe any technology can overcome a lack of understanding in math, physics, or statistics (I am wary of the term ‘citizen data scientist’), I do believe these technologies can help organizations overcome the engineering/developer talent shortages.
Data science technologies, through their GUIs and workflows, make it easier for analytics groups to develop production solutions. This is especially true when it comes to stringing together data engineering work with data science modeling. Data science technologies allow both activities to occur in a single, often web-based platform.
Without these technologies, I have witnessed data science groups develop prototypes that are often sent to a different, production group to turn into a production solution. This leads to significant delays in deploying solutions and has additional costs associated with funding two separate groups aimed at accomplishing the same task. Introducing data science technologies that make it easy to develop production solutions saves money in the long run.
Without a data science technology, analytics groups are often forced to use a development process that requires them to adhere to developer-oriented tasks. Specifically, without a data science technology, analytics groups must develop, commit, and version code in a traditional code repository. While some analytics groups have the engineering acumen, many analytics groups are not developers. Finding a data science technology that is web-based, and has these features hidden behind a project-like platform, prevents organizations from training their analytics group on developer best-practices and opens the door to who can contribute to analytic solutions.
Off-the-shelf tools can be feature-rich but often lead to lock-in and more constraints than benefits. The upside is that there are standard, open technologies such as SQL, R, and Python that support a large extent of data science needs. Open technologies, rather than proprietary languages and black-box functionality, create advantages in recruiting, increase the efficiency of onboarding, and greatly improve portability of solutions. That said, without a platform for harnessing the collaborative efforts of a team, standards can be hard to manage. Selecting a robust engineering process and data science toolset allows analytics groups to standardize a rapid development method that maximizes productivity and value.
Selecting the right technology
When evaluating data science technologies, it is important to consider the skill sets within your analytic organization. If you are lucky enough to have an analytics group with significant engineering or developer skills, you have more options into which technologies, if any, are required to facilitate efficient development and deployment. If you’re like many analytics groups, with little or no engineering or developer skills, introducing a data science technology into your analytic platform may be a wise decision. The cost of certain data science technology solutions may not seem as high when you consider the additional talent you need to hire to offset the lack of engineering or developer skills.
At WebbMason Analytics we have the experience, flexibility and multidisciplinary analytics experts to guide our clients on the planning, implementation, and deployment of data science projects. Let’s discuss how we can help your organization.