Mango share some keys to Data Science readiness

Following on from our series on programming languages, I’m delighted to introduce a guest post on Data Science readiness.

Having the strategy or aspiration to make use of Data Science is one thing, executing it effectively is another. Too many businesses I know have hired data scientists without really knowing what to do with them or what they need to succeed.

So, I’m pleased to introduce a new guest blogger. Mark Sellors is Head of Data Engineering for Mango Solutions.

It helps to hear a Data Engineer’s perspective, as that can be such a foundational role to a successful Data Science team.

Mango Solutions provide advanced data analysis solutions, consulting, training, and application development for some of the largest companies in the world. They offer a number of bespoke products for data science, including validation of open-source software for regulated industries. I previously mentioned the R programming expertise of Richard and his team at Mango.

Anyway, without further ado, over to Mark to share how you can ensure your Data Science readiness…

The keys to Data Science readiness

Any business can use Data Science to improve decision-making, but success, and long-term success in particular, is often elusive. Data Science is not a magic bullet, cure-all solution for every problem your organisation faces. However, when done properly, Data Science provides a rich and powerful support for corporate decision-making.

With this in mind, having a Data Science team – a group tasked with improving corporate decision-making through the use of advanced analytical techniques – has become increasingly mainstream in the modern enterprise. While the internet is awash with success stories, many organisations’ Data Science efforts fail to live up to expectation. In this post we’ll take a look at the biggest strategic driver of success in this field and how your business can realise the maximum potential of implementing a Data Science team/capability.

Finding the right people

The current market for Data Scientists only exacerbates the problems with implementing a successful Data Science capability. When demand for a particular type of employee outstrips supply – as it does for Data Scientists at the moment – it’s increasingly challenging to find the right people. A hastily implemented Data Science team, staffed by the wrong people stands little chance of delivering on the promise of using data to significantly enhance your decision-making.

For example, many organisations looking to repeat the successes they read about, immediately make the jump to hiring PhD level statisticians, often straight out of academia. These poor individuals are often thrown straight into the organisation, with little understanding of how and where their skills might be used, without access to fundamental data sets, and with ineffective tooling. Though this approach is widespread, it rarely brings with it the sorts of rewards that were originally envisaged and, in fact, more often ends in disaster.

How do companies succeed where so many fail?

Really well-implemented Data Science can seem almost magical as they can deliver fresh insights and help steer the business towards ever greater successes. However, like most good things in life, it’s not something that comes without effort. The amount of effort required varies wildly from one organisation to the next, but effective planning prior to on-boarding your first data scientist is essential to ensure they’re able to hit the ground running.

1) Provide access to data

First and foremost, those plans should include arrangements for access to data, as without that, your new Data Scientists will be stuck twiddling their thumbs. Good Data Scientists don’t come cheap, and the last thing you want to do is to be paying them to surf the web while they wait for access to key data to be granted. Get out in front, and make sure that’s in place for day one.

2) Make sure the right tools are available

You can’t dig a building’s foundations with a spoon, there are proper tools for that job. Data Science is no different, so it’s important to provide appropriate software tools to ensure your Data Scientists can work effectively and efficiently. To some extent this may be informed by location of the data. Most commonly this will take the form of some combination of an SQL client along with R and/or Python, but some organisations may require more esoteric tooling, as the existing organisational data management practices dictate.

3) An empowered team, is a successful team

Perhaps more important even than effective planning, is empowering your team. It’s unlikely a new Data Science team would have the essential business domain knowledge to fully support the organisation immediately, but they should be empowered to engage with appropriate internal domain knowledge expertise where necessary. It is essential when approaching a new Data Science project that the full problem space is fully understood in order to find the most effective response to any given question.

4) Keep your IT team in the loop

This empowerment also needs to extend to areas like software tooling and computing resources. Here, it’s often useful to have a close relationship with your existing IT team to ensure that new requirements are met in a timely fashion. Alternatively, it’s worth considering the engagement of dedicated specific resources in the form of data engineers and architects, who will be able to help with moving, organising and storing data.

5) Know what you want to achieve

This should probably be higher on the list, but I thought it would stay in mind at the end. To really make the most of your Data Science team you need to have an idea of the sorts of questions you’d like to have answered. Take advice on whether these questions are answerable given the data you have access to. Use this information to put together a plan for your incoming team. In fact, use the plan as an aide in your recruitment process.

There is no one right way to develop a Data Science capability and no one-size-fits-all definition of what a Data Science team looks like. However, by laying the strategic groundwork up-front, and ensuring your business is ready, your Data Science initiative has the best possible start, and therefore, the highest possible chance of helping your business achieve its goals.

Which tips will help you achieve Data Science readiness?

Thanks for Mark for his sensible advice. The number of links I was able to add to his post, shows how many times our thinking clearly aligns. I hope that post helps you ensure your efforts, in securing investment in a Data Science team, do not go to waste.

Which key or tip resonates with you? Any that are particularly relevant for your business or echo your experience.

Do share your thoughts in comment boxes below or on social media. Happy to explore the most important topics in future posts.