From 2018 to 2019 the demand for Data Scientists rose by 56%. This rise in demand is only expected to increase in 2020 and beyond. It is clear that the supply of Data Scientists is unable to keep up with global demand.

As organisations create more and more data, the need to understand and provide business value from this data will be a key differentiator. The challenge organisations now face is finding the right data science skills to do this. The cost of a Data Scientist is rising with a study from PromptCloud showing a rise of 6.43% in average salary in 2018 and as costs rise so does the difficulty in demonstrating business value.

One of the key challenges for organisations investing in data science capability is the need to provide a return on investment. Alongside the growing costs of Data Scientists are the challenges that organisations face in deploying data science to deliver value. As much as 80% of data science fails to make it into production. When this figure is paired with the rising costs of Data Scientists the ability to generate a return on investment becomes a significant challenge facing the industry.

 

Overcoming the Data Scientist shortage

Other than training data scientists inhouse or partnering with specialist providers of skilled resource the only approach for most organisations is to look at how they get more value from the data science team they currently have. How do you make your data scientists more efficient and focused on improving the business value generated by getting more data science into production.

Data Scientists only spend around 20% of their time doing data science, with the rest made up with data access, data manipulation, data cleansing, model re-coding, systems integration and so on. Google estimate that “mature machine learning systems are typically made up of (at most) 5% machine learning code and (at least) 95% glue code”. This means you are paying your Data Scientists to spend between 80-95% of their time not doing data science.

The key question is this: how can you ensure that some of the most valuable resources in your organisation are supported to deliver real value, consistently and efficiently?

Increasing the impact of your data science

Some organisations have invested in a deployment team or ModelOps team to help them deliver more data science. But a key opportunity for organisations is to improve deployment via automation. By automating the deployment of your data science models, not only can your Data Scientists focus on data science, but you will also increase the number of models that make it into production and your efforts will scale without the need for large-scale recruitment. Removing these deployment challenges will create huge productivity increases across your data science team while helping to deliver more data science models to production. All of this contributes to the delivery of business value from data science.

 

Deploying data science with Lumen

Lumen from Bellrock Technology is the world’s most complete data science deployment platform. Lumen provides a range of capability to allow you to easily and efficiently deploy data science models into production and start generating insight to users. Lumen automatically containerises models and builds data pipelines taking away much of the glue code currently created by your data science teams.

Find out more about Lumen in our Lumen product guide.