View all newsletters
Receive our newsletter - data, insights and analysis delivered to you
  1. Technology
  2. Emerging Technology
May 5, 2019updated 07 May 2019 11:41am

Can Machine Learning Be “Productised” for Business Problem Owners?

Good news: the answer is "yes", argues Nathan Korda, Mind Foundry's director of research

By CBR Staff Writer

Data is an ever-growing asset in businesses across all industries, and with increased data volumes comes a wealth of potential insights.

Business problem owners know data can help them unlock business value, but they often lack the skill set to create appropriate data-driven solutions, writes Nathan Korda, Director of Research, at Mind Foundry.

This is where accessible machine learning comes in. Far from being the preserve of expert data scientists, machine learning has matured, and is ready to be consumed by your company’s existing employees. The next generation of data science platforms will turn data science into an easily deployable business tool, transforming business problem owners into ‘citizen data scientists’.

Accessible Machine Learning: “Even Data Science Teams are Unable to Service all Parts of Their Business”

Over the last decade, the major success stories for machine learning in business have come from companies with vast budgets and large teams of data science experts. However, even within these large firms, data science teams are unable to service all parts of their business, and the data science sector as a whole is suffering from a shortage of skills.

Nathan Korda, Director of Research, Mind Foundry.

A recent KPMG CIO Survey found that big data and analytics is the number one place of need for new recruits. Nearly half of the CIOs who participated in the survey said they suffered from a big data and analytics shortage, with 38 percent reporting a shortage in AI skills.

The good news is that data science can be productised so that anyone who is proficient with a spreadsheet can create data-driven solutions to their problems.

In many businesses, data scientists form working groups with business problem owners to deliver solutions to specific problems. This structure does not solve the disconnect between employees that understand the analytics, and employees that understand the business value.

Content from our partners
Green for go: Transforming trade in the UK
Manufacturers are switching to personalised customer experience amid fierce competition
How many ends in end-to-end service orchestration?

In contrast, using an accessible machine learning platform, with some simple clicks, business problem owners themselves will be able to explore the insights buried in their data, and both deploying and managing decision-making algorithms.

We’re Talking in Hours rather than Months

Data science is traditionally labour intensive. Approximately 80% of a data scientist’s workload is consumed by data preparation. Model tuning and visualisation often take up the remaining time allocated to any project. This leaves little time dedicated to the business problem itself.

An accessible machine learning platform offers a multiple boost to this process. It will guide any user through the data science process from start to finish, suggesting actions to clean and to correctly format data, as well as recommending the most suitable model for a particular problem and data set. It will also allow the business problem owner to directly and quickly input their contextual knowledge and intuitions about the business problem at hand which are not present in the data. This means that more appropriate solutions are found for more valuable problems more quickly.

Humans and Machine Learning Platforms as Colleagues: A Symbiotic Relationship

Despite the undeniable power of data-driven analytics, the business problem owner remains key to the successful deployment of this technology. Machine learning platforms and humans have complementary skill sets. Properly applied, machine learning produces unbiased, data-driven judgements, effective risk management and insight that is beyond human capabilities. However, any knowledge that is not contained in the data is out of reach of machine learning. That includes the human’s ability to contextualise data for the business problem at hand and to set the goals of the machine learning analysis appropriately.

Take the example where error codes are present in a large data set collected from sensors. A machine learning platform will struggle to understand these error codes, and may interpret them incorrectly, leading to poor modelling performance and minimal insight. A human who is close to the business process can quickly provide an explanation, such as sensors being out of range, and even enable the platform to raise flags and mitigate problems arising when these error codes are recorded.

Machine learning reduces the time involved in analysing data and enables the analysis of large amounts of data, but the user should be the one directing and controlling the analysis and its application. Accessible machine learning augments the human skill sets, taking on the laborious data and model delivery tasks and letting the problem owner concentrate on the business problem itself.

Looking further forward, accessible machine learning platforms will help with knowledge transfer between one employee and the next. Imagine if a problem owner or data scientist deploys a machine learning model to optimise and monitor a specific business process but then moves elsewhere. Rather than getting lost in a sea of deployed scripts, those models will still exist within, and be made accountable by the machine learning platform. This will allow new users to pick up where their predecessor left off, without disruption or the need for extensive technical training.

Moreover, accessible machine learning platforms will help data scientists as well. While they are currently stuck performing all the labour-intensive work that bears little or no relevance to any business value, using an accessible machine learning platform will enable them to get closer to the business value and gain a deeper understanding of their company’s needs. Equally, such platforms will free up data scientists to work on truly new and innovative technology.

Complex Business Problems – Simple Solutions

Machine learning platforms can help business problem owners across a range of industries. Take, for example, a telecoms provider aiming to reduce customer churn. Logging every single customer behaviour creates a vast and sprawling data set – making it a huge task to even assemble and prepare data before any analysis can be run. With the business problem owner guiding the way, a machine learning platform can be used to prepare data and then suggest and deploy models to identify customer personas that are likely to rapidly open and close accounts – in minutes, not days.

In a different scenario, imagine a manufacturer experimenting with new materials and chemical compounds. Testing these materials is expensive and time-consuming, as each trial carries a high cost in raw materials and expert time. With the support of an easily deployable machine learning platform the problem owner at that manufacturer can predict the most valuable experiments to run – those that will deliver the highest information gain. Going forward the manufacturer can explore what data is important to collect, what can be discarded, and what should be closely monitored – achieving huge cost savings on expensive sensors and data storage.

Levelling the Machine Learning Playing Field

With accessible machine learning platforms, human intelligence is enriched and augmented, and machine learning capabilities become available to anyone with access to data. Using data-derived insights, newly empowered ‘citizen data scientists’ can make better decisions and cope with the ever-growing volume of data in a profound and principled way to make transformations at every level of their business.

Open Source Platform Aims to Democratise “Machine Learning Zoo”

Websites in our network
Select and enter your corporate email address Tech Monitor's research, insight and analysis examines the frontiers of digital transformation to help tech leaders navigate the future. Our Changelog newsletter delivers our best work to your inbox every week.
  • CIO
  • CTO
  • CISO
  • CSO
  • CFO
  • CDO
  • CEO
  • Architect Founder
  • MD
  • Director
  • Manager
  • Other
Visit our privacy policy for more information about our services, how New Statesman Media Group may use, process and share your personal data, including information on your rights in respect of your personal data and how you can unsubscribe from future marketing communications. Our services are intended for corporate subscribers and you warrant that the email address submitted is your corporate email address.