Our Data & Analytics team explores key features, business benefits, industry applications, and technical capabilities of Microsoft Cortana Intelligence Suite.
Blog three of a series.
We believe there is a global shift occurring from business intelligence to predictive analytics.
At the forefront of that revolution: Microsoft Cortana Intelligence Suite – a fully managed, subscription-based analytics toolset – that allows organizations to challenge traditional, often flawed approaches to predictive analytics.
Cortana empowers technologists and business professionals to maximize their talents to help predict the future through data science.
The Dominance of Predictive Analytics
In just the past two years, the nature of our clients’ needs have dramatically changed.
Predictive analytics has clearly moved to the forefront of the corporate stage.
Clients used to ask us: “How can we give people better access to trusted information?”
Now, clients ask: “How can we predict the future” to deliver better customer value, secure market position, create new products and services.
Attempting to Predict Analytics with Azure Data Lake
Data Lake is the most recent and common approach to attempting predictive analytics. It looks something like this: Set up a Data Lake to ingest all of your corporate data and hire an army of data scientists to find business opportunities.
Problem is that this approach is flawed. Big data requires extensive administrative and developer manpower. Data scientists are ill-equipped to differentiate between anomalous “black swan” events and the levers that can change business outcomes.
However, when business leaders and employees look at the same data, they can usually tease out the differences. Because they know the backstory of what actually happened in the business.
- “Finance changed how they handle accruals in the current fiscal year” or
- “Sales decreased because of a supply chain disruption” or
- “The increased return rates were caused by a manufacturer product recall”
Those nuances are critical. They happen in every company. If analysts fail to take that business context into account, data science can lead to misleading results.
Traditional Data Science Model
The traditional data science model espoused over the past decade goes something like this:
A company maintains a pool of data scientists, who proactively scour data – both internal company data and external data – looking for opportunities. The opportunities may drive up operational efficiency, reduce risk, or grant a competitive edge. Their belief is that, while data scientists are expensive resources, the benefits will pay off in the long-run.
In global financial markets, “quants” – or quantitative financial analysts – introduced advanced modeling techniques to predict market performance. They have changed the face of the equities markets, not only increasing returns, but introducing ever more sophisticated algorithmic trading systems, such as high-frequency trading platforms.
In a recent Forbes article, The Quants Are Taking Over Wall Street, Paul Tudor Jones, a billionaire hedge fund manager, declared that “…the era of complex math and computer-automated algorithms ruling Wall Street…is well on its way.”
He went on to say: “No man is better than a machine”… “And no machine is better than a man with a machine.” In other words, it is knowledgeable people with sophisticated machines that drive up value.
The “quant” examples shows how the traditional data science model can successfully work. However, believing that this example applies to corporate America is misguided. Equity markets are highly regulated. Data is consistently available and the levers to pull are well defined: buy, sell, puts, calls, shorts, etc.
But, insurance, banking, healthcare, manufacturing and other industries are far more complex. Their data is more sophisticated, less clean, and subject to wide interpretation.
These industries are impervious to the traditional data science model. Indeed, at Centric we have seen multiple organizations try-and-fail at the traditional data science model.
Most industries demand a different approach.
The New Data Science Model: Combining Business, IT, and Analytics to Create Data Science Solutions
When a traditional data scientist performs her job, she must demonstrate three main skills:
- Data Acquisition: Finding, aggregating and cleansing, and restructuring data.
- Predictive Modeling: Building, training and tuning predictive models.
- Business Analysis: Iteratively researching business scenarios, adjusting models and incorporating feedback from implemented business changes.
The New Data Science Model acknowledges the obvious fact that one person cannot excel at all of these skills in a complex business environment. Highly intelligent technical people may still lack business knowledge, interpretation of data anomalies, and organizational influence necessary to implement change.
Without implementing their ideas, it’s nearly impossible to get the feedback required to refine predictive models.
The new approach calls for collaboration across multiple parts of the organization with each person bringing their strength and talents.
- Put Data Acquisition in the hands of IT: They are highly prepared for data management responsibilities, including integration, cleansing and governance.
- Put Business Analysis in the hands of the business: No one is better prepared with the institutional knowledge to understand how historical business operations can skew and influence underlying data. They know business priorities and can exert influence (and budget) needed for investment and experimentation.
- Keep Predictive Modeling in the hands of analysts: Those who are skilled in statistics and predictive modeling. They do it well and they love it. The analyst’s role is to request information from IT as well as partner with business stakeholders.
Those three functions serve as the core of “data science.” Bridging the business, IT, and analytics teams can help achieve a balanced solution. Transitioning to this new approach remains a modern challenge for companies.
Microsoft Cortana Intelligence Suite has emerged as the answer to this challenge. Their cloud platform is tightly integrated. More importantly, Cortana interconnects tools that allow each individual to efficiently perform their work.
The output can be shared with colleagues in different roles. In summary, Cortana is all of these wrapped into one: an integration platform, a powerful computational engine, and a data collaboration environment.