<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=1691976594387096&amp;ev=PageView&amp;noscript=1">

Building a Data Science Organization



Data Science is becoming more and more widespread every day. Organizations everywhere are realizing the positive potential it has and are quickly working to join the masses.

The biggest challenge these organizations are facing however, is finding people with the right skills. Many organizations start with a single data scientist, which can lead to two different negative outcomes. One, the scientist doesn’t have a team or the tools they need in order to succeed, or two, the scientist has expertise in one area, but no knowledge in other domains within data science.

Alternatively, finding one single data scientist with all the key skills they need can also prove difficult and expensive. To overcome these challenges, there are four potential strategies organizations can adopt which are outlined here.

1. Data Science as a Service (DSaaS)

Especially when there are time and budget constraints, having data science related work completed by outside sources can be worth looking into. This approach can also be useful with an organization’s first data science project where, rather than investing time, money, and resources into forming an internal team, it is important to first determine how data science fits into the organization’s specific goals.

DSaaS can take two different forms:

  • One is hiring a team of consultants. The number of consulting companies is growing, which makes it easier to search for a company with expertise in what the organization needs and makes this approach more likely to be successful. While consultants working in the organization’s industry may not have specific experience in all the systems and data the organization uses, they will most likely be familiar with the types of problems and common issues that the industry faces.
  • The other is using available tools and solutions that help with data analysis. If there are people in the organization that have some knowledge of data science, this can be a great low-cost solution that is easy to understand and implement.

2. Crowdsourcing

In 2009, Netflix held a competition in which they asked data scientists to predict the probability of someone enjoying a movie based on recommendations that get served up through their past activity and preferences. Over 50,000 people signed up to compete for the $1,000,000 prize. Instead of hiring and paying a team for over a year to solve this, Netflix ended up with 50,000 distinct solutions to this problem.

Companies such as Kaggle and DrivenData have since launched several platforms and hosted hundreds of data science competitions. These bring together the best minds and talent to solve complex problems. There are numerous benefits to these competitions:

  • Many people working on the same problem generating a broad range of ideas
  • Countless different techniques that could not all be funded by a single organization
  • Increasing quality and efficiency of solutions through competition, keeping participants motivated
  • Positive public relations (draws large range of intelligent people and helps hiring)
  • Collaboration

While the possibilities are endless, when running such a competition, the organization must state the problem very clearly, and provide a very clean dataset in order to avoid potential risks. Many a times, the solutions can be extremely complex, which can lead to very high implementation costs.

Overall, crowdsourcing has no guarantees and can pose some data privacy concerns, but it can also deliver a wide range of creative solutions.

3. Rotation and reposition within your organization

The option of repositioning current members within the organization is often overlooked. While the organization may not have any existing employees that work in data science, that doesn’t mean no one in the organization has the required skills. It is likely that there are at least a few people who have some technical know-how, such as knowing how to obtain, presenting, or modeling data. It can be in an organization’s best interest to help those members expand their skills, as they already know the organization, the data, and its problems.

There are several different courses of action that can be taken to go about expanding skills of existing employees:

  • Start by researching with books and the internet. Books can help provide a plan for training and the internet can provide more specifics for the situation/problem the company is trying to solve.
  • Next, looking into different organizations that provide data science training can also be of value. These courses range from a few days to a few months, and can be more generic, or specific to a product or tool. Starting with the general courses is the best option. Once complete, only then moving into more specific training can help avoid relying on one tool for all of the organization’s issues. Some credible sources for training include:
  • Lastly, universities are constantly creating and offering new programs, ranging from business analytics to data science. Certificate and degree programs are both offered online and on campus. Because all the programs are different, finding a program that fills the skills that the organization currently needs is crucial.

4. Build a team from ground up

The other common strategy is to build an entirely new team from ground up. There are two important conditions to this option working successfully: 1. the organization might find the right people to hire, but their domain knowledge may not be a must-have for the business. 2. It can be difficult to find the right people, and new people means they don’t have deep domain knowledge. If deep domain knowledge isn’t critical, this approach can work well. Data science people are also in very high demand, and studies like those from McKinsey & Company show that there is a shortage of available talent.

Data science is still very new, and to be successful it Is essential to find the right people for the job at your organization. While each of the people strategies highlighted here have its own benefits and challenges, they are an excellent place to start identifying what will work best for you and your organization.

To learn more about the possibilities of AI and data science, visit The Spur Group’s client stories.

Clay Campbell

Clay Campbell

Clay Campbell is a Client Delivery Director in the Data and AI practice at The Spur Group. Clay has spent over a decade working with leading companies on how to utilize technology and tools for the advancement and automation of the world. As a well-established technology and thought leader in Applied AI, he has made it a priority throughout his career to apply current technologies as well as new techniques to develop elegant and creative technical solutions across all project phases.