Predictive analytics. Data mining. Machine learning. Forecasting. Text Analytics. Optimization. Are you starting to hear more about advanced analytics these days? Are you trying to get past the hype and figure out what’s real and what’s not? Are you still wondering how you can use analytics to improve your business results?
Research firm Hurwitz & Associates surveyed and interviewed people across all industries to better understand which tools they are using to find hidden patterns in data. Hurwitz has published “Advanced Analytics – Hurwitz Victory Index Report,” a good reference on the concepts, tools and players in the advanced analytics market. I want to highlight three sections and topics that have been resonating with our customers.
Customer examples and real-world use cases
First, not everybody knows what we mean by “advanced analytics.” Like many business and technology terms, it makes more sense when you give real-world examples. The Hurwitz report (p.13) calls out four use cases for advanced analytics that make the term much easier to grasp:
- Predicting consumer behavior – Everybody wants to be able to do this, of course, and when you have enough data and the right tools, you can find hidden patterns in the data. With these patterns you can personalize offers, identify your best customers, anticipate their needs, and even find unhappy customers. As a result, consumers get more-targeted and better offers, and you get happier customers who spend more.
- Forecasting sales and inventory – Eliminate much of the guessing that goes into your sales and forecasts. With a large enough body of data, a company using advanced analytics can improve the accuracy of its sales models and reduce inventory. By understanding different factors that can affect your forecasts – for example, holidays, back to school – you can ensure that you have the right products in the right store locations at the right times for the right customers.
- Anticipating failures in machinery – In machines as varied as medical devices, elevators, airplanes and earth moving equipment, sensors can stream data to manufacturers in near-real time. By using advanced analytics, you can dispatch service reps to perform preventive maintenance on the equipment that is most likely to fail. You can combine structured and unstructured data like technical notes, Twitter feeds and Facebook posts and apply warranty analytics to take preemptive action on quality problems.
- Detecting anomalies and reducing fraud – The only practical way to improve detection in fields like insurance, Medicare/Medicaid, lending and credit is with analytic tools that find suspicious patterns in the fire hose of incoming data. Think about this the next time your credit card company phones and asks you, “Did you just purchase a one-way ticket to Fiji?” This application of advanced analytics saves you time and your bank money.
So the technology angle on advanced analytics includes the algorithms for finding patterns and the visualization tools to display them as charts and graphs. The business angle is how you make decisions based on them.
3 important trends
The Hurwitz report identifies 11 trends (p.8), and I want to emphasize three:
- Accessibility – Like PCs in the 1980s and the Web in the 1990s, analytics in the 2010s are valuable for what business users can do with them without having to learn a completely new skill set. The winners in the category will be the products that are flexible and extensible for data scientists yet easy-to-use and accessible to business users and managers.
- Visualization – A big part of that accessibility is the presentation of data. Advanced visualization tools reveal patterns that would take the human eye too long to find in data tables, spreadsheets or basic charts. The Hurwitz report points out that visualization tools in analytics products go a long way toward bridging the gap between business users and data scientists.
- Programming Languages – Spend any time looking into advanced analytics and you’ll come across discussion of open source technologies like R, Python and PMML. R is an open source programming language favored by academics and statisticians. Python is more general-purpose and used by data scientists. PMML, or Predictive Model Markup Language, is a standard for deploying predictive models on many different database platforms that analyze changing data in real time.
The role of the data scientist
In large enterprises, data scientists are developing complex models with high-end analytic software and making the biggest strides in the use cases I described above. Because the tool landscape is still far from one-size-fits-all, many data scientists prefer to maximize their effectiveness and productivity by working with a particular vendor.
The data scientists interviewed for the Hurwitz report (p.6) understand how to spot patterns in large amounts of data that point to fraud, market trends and customer preferences. Of course, in spite of how much they enjoy swimming in the analytics, the best ones never lose sight of the interests of business users.
However, not every organization can afford data scientists, and that underscores the need for easy-to-use, accessible tools that put model development and analysis in the hands of more business analysts and managers.
Whether your organization is already basing decisions on advanced analytics or still gauging the fit, you’ll find the report useful. If you’re already using Dell Statistica, you’ll be pleased to see that Hurwitz ranked it a high leader in Go-To-Market Strength and a victor in Customer Experience Strength. (Don’t you love it when an independent survey validates your buying decision?) In fact, Dell Statistica received the highest score for value compared to price.
Have a look at “Advanced Analytics – Hurwitz Victory Index Report” and start planning your next move. Send me any questions in the comments below and let me know what other content you’d like to see from Dell around advanced analytics.