Driving breakthrough innovation and change
To achieve differentiation and speed to market, businesses need to innovate and drive change.
Speed to market and differentiation are two key factors for business success. To achieve both, organizations need to rapidly innovate and drive change. Development organizations, in particular, need to consider how to survive in today’s global, turbulent economy and learn to leverage data science and analytics in order to drive innovation.
Only a few years ago, development organizations built and delivered products based on a version-by-version and release-by-release process. All changes would be aggregated together, involving lots of unit, integration and acceptance testing as one big chunk of code. Customers would then dare to install the code and test within their own environments. The sheer amount of change or added capabilities from release to release presented a daunting task for many clients, and customer feedback would have to be rolled into the next fix pack or release.
To combat this, development organizations are adopting an agile philosophy of constant learning and innovation, instilling quality in everything they do. This approach, combined with cultural change, helps empower teams to achieve what once seemed impossible. Using data science and analytics is key to this success, as artificial intelligence (AI) and machine learning have become the foundation for deriving the next generation of intelligent solutions. Fail fast learn fast, reiterate and improve.
Agile development, as shown below in Figure 1 from the IBM ebook Agile for Dummies, requires commitment across the organization and beyond.
Figure 1: Agile development requires cultural change across an organization.
Right tools for the job
The design and interface for any tool (and its usability) — whether it’s a home appliance, power tool or piece of software — must be intuitive and simple to use, while also being fully functional in achieving its outcome. Data science tools must enable different personas in a project to collaborate using the collective skills of the team to define, create, test deploy and manage a machine learning application. Data engineers, data scientists, researchers, machine learning engineers and production engineers all must be able to interact with each other and the assets each builds — whether using Notebooks or graphical interfaces that guide less experienced users through automated step-by-step processes. Vendors’ solutions based on open source and their commitment and contribution to the community helps promote a common code base and interoperability. IBM Data Science Experience (DSX) and machine learning solutions have been built around this very philosophy, as shown below in Figure 2.
Figure 2 – Right tools for the jobs
At IBM, for example, the core analytics development team adopted an agile mindset in its development and operational processes. Their customer support and service application uses IBM DSX and machine learning algorithms to evaluate faults raised by customers, compare them with similar issues, evaluate issue durations, track those engaged in resolving faults, monitor customer comments and sentiment and discern how these variables impact IBM key metrics and satisfaction ratings.
This process helps us identify existing and potential quality issues. As a result, support staff can more rapidly and accurately identify issues, pinpoint their causes and find solutions. The results are higher Net Promotor Scores and lower detractors. Think of agile as a continuous journey of improvement and excellence — less so a destination.
Execute, execute, execute
In sports, we hear about blocking and tackling — seeing a risk, then stopping and removing it to pave a clear run to the touchdown. Software development involves a similar approach. We identify what impedes our progress and find an alternate route that can lead to success. Prototyping in small groups can provide a fast and efficient way to produce code that can be tested against various hypotheses and refined through an iterative process.
To help customers consume smaller, well-tested code that’s designed to be less invasive and problematic for production systems, we think it’s important to clearly define minimal viable products (MVP) and instill a maniacal focus on quality while moving toward a continual release cycle. This process fits nicely with the continuous delivery approach, as everything moves through an iterative process. Say goodbye to the days of big general availability (GA) code drops.
Execution also includes the bigger ecosystem. Machine learning is heavily data science focused. The IBM Machine Learning Hubs are centers of expertise around the world where customers can meet experienced IBM staff to work through how to best implement an idea and progress it toward production, using some of the principles discussed above. These Hubs also provide education, guidance and knowledge transfer — all of which can help build long-term client relationships.
Machine learning use cases
According to this article that references an October 2016 Nilson Report, global credit card and debit card fraud resulted in losses amounting to $21.84 billion during 2015. Machine learning can score each transaction with degrees of confidence to help authorize a transaction or decline it due to suspected fraud, alerting both the customer and financial institution in real time. Without this level of real-time machine learning intervention, loopholes in transaction policies could be exploited by cyber criminals, potentially without humans finding out for months.
In closing, driving breakthrough innovation and change requires a cultural shift that empowers teams and individuals to dream big and reach for the impossible. It’s about rapid prototyping, blocking, tackling, agility, collaboration, keeping things simple and instilling quality in everything you do. It’s also about taking those internal principles and applying them to how you engage your customers, building long-term relationships.
Take the next step and contact the skilled professionals at one of our IBM Machine Learning Hubs for help bringing your machine learning projects to fruition.