Guidance and results for the data rich, yet information poor.

Category: Articles

The 5 Biggest Lies About Data Mining

biggestliesMore and more people are catching on to the idea that there is more to data mining than the “big data hype” would suggest. But there are still 5 very big, persistent lies (or myths, if you prefer) that plague conversations about data mining or predictive analytics.

If your organization can move past these lies you will be able to reap the benefits of data mining and predictive analytics as never before. Each lie holds you back from doing all you can, so it’s important to understand them and to debunk them, one by one.

1. All Companies Use Data the Same Way

A lot of people think “Amazon” or “Netflix” when they think about predictive analytics. It’s certainly true that predictive analytics can be used to offer individual recommendations to customers, and it is true that in certain industries that capability can increase sales.

But there are many other ways to apply data mining and predictive analytics. If you want to increase sales you can also use it to place products in an advantageous way. You can use it to adjust the wording or the timing of a sales letter to get a very profitable incremental improvement. If you want to decrease waste you can use it to decide exactly how many widgets to order. If you want to decrease fraud you can use its ability to target useful outliers to help you find the fraudulent and bring them to justice.

Every organization has different needs, different, pressing problems, and even different stores of data. All of that will help determine what the best and most profitable projects will be for your company.

2. Choosing the Right Software Package is the Most Important Step

Many organizations approach data mining backwards. They say, “Well, data mining is a big thing now. We should probably be doing some of that.” Then they send someone off to decide which software package to buy.

The result is an expensive purchase and an expensive implementation that goes absolutely nowhere.

Instead, the company should say something like this: “We really need to figure out why sales plummet in April and what we can do about it. Data analytics can help with that.”

The company can then turn to one of two resources to solve this business problem: any open-source data mining software, which is free, or Microsoft Excel, which companies already have. From there it’s a matter of preparing the data and setting up the project in a way that ensures that it is capable of actually asking the question.

3. You Need to Hire a Data Scientist

TMA has covered this misconception before. A data scientist is usually the last person you need.

Data scientists create algorithms. Many of those algorithms already exist inside of Excel or other programs. 90% of the time nobody needs a better algorithm. They need a better approach for using the algorithms they already have.

Does that mean there’s no value in developing more powerful algorithms? No, of course not. That’s like saying there’s n o value in creating a better truck. Most companies, however, don’t directly hire engineers to create better trucks. They buy the best trucks on the market, then look for qualified people to drive those trucks.

4. Data Mining is Only for Huge Organizations

Organizations of any shape and size can use data mining to some degree, depending on what data they are collecting and what they are trying to do.

Obviously size and budget do open up the number of tools you can buy. If you absolutely think it’s necessary to track the movement of shoppers around your store to solve your business problem then it helps to have the budget to buy the software and hardware that can do that. Often, however, shopping patterns can as easily be determined by digging into the sales records of any store.

Data mining is most often effective at creating incremental change. Small changes can often produce big results, and small changes can stack to create even bigger results. A 1% increase in the effectiveness of your sales letter campaign doesn’t sound like much until you think about just how many people you’re mailing, just how many new customers that 1% increase would represent and just how much more revenue will be generated. What would happen if you could increase that number by 1% again? Big things.

It’s not the size of the organization but the needs of the organization that matter.

5. You’ll need four more years of training to ever make this happen.

Training is absolutely key to learning how to use data mining correctly. However, it hardly requires anyone in your organization to go pursue a second 4 year degree. A few weeks of seminar training are often all that is necessary to fully realize the potential that data mining and predictive analytics hold for your organization.

Ready to get started? Register for TMA’s free webinar today, or sign up for any one of TMA’s more intensive training courses.

Columbus Zoo: Doing Data Right

ZooIt’s one thing to talk of actionable business intelligence in the abstract. But it’s also great to spot an example of an organization that is truly approaching their data in a strategic way.

Data Informed ran an extensive interview with executives at Columbia Zoo to talk about some of the focused data analytics projects that they’ve already developed and run at the zoo. Some early successes include:

The zoo using BI to determine whether or not new exhibits are likely to be successful.

Using data analytics to make decisions about under-performing food and retail stands throughout the park, reducing waste and increasing profits.

Using data analytics to make an effort to foster more effective interactions with the donor base.

Of course, TMA has shared many data success stories on this blog. But this interview was very telling as it covered the thought processes behind the success stories, and that is worth highlighting.

For example:

We started to look at what we didn’t know about ourselves…and said, “We have data problems. We don’t really know enough to be efficient at what we are doing or to make really good, fact-based decisions.

Or:

We had a couple of key projects that we knew could make an impact right away…We knew we had a few targets we could fix right away and make a big impact with, and that in itself gave us a chance to kind of snowball the system in to where other departments looked and said, “Oh that’s what you can do with that, and now we can visualize that and see that, we can touch these reports. And, by the way, can we have this and this and this and this?”

This process of discovering those first key projects and targets are really key to creating a truly data-driven organization. Do what matters most first. Get successful, create results, and move on to the next one. Waste zero time on hype–don’t create new problems trying to solve old ones. This is excellent strategic thinking at work.

If you’d like to learn how to develop this kind of business intelligence strategy for your own business then it’s time to think about laying the groundwork that will put you in the right mindset to do just that. Get started today by registering for TMA’s free webinar, Data Mining: Failure to Launch. By the time you’re done you should be well on your way towards identifying those key targets and issues in your business environment.

Prescriptive Analytics: The Next Step in Business Intelligence

prescriptive-analyticsThis month, the subject of prescriptive analytics came up in the Q&A that ends each one of TMA’s free webinars. But just what is “prescriptive” analytics?

Prescriptive analytics is predictive analytics taken to the next logical level. Instead of merely predicting the future it asks how a business might start treating specific issues based on the model. This is a computer-assisted task: one that helps business intelligence analysts choose from perhaps dozens of options. The question now becomes, “out all of these good options, which one will deliver the best results the fastest?”

The Analytic Journey,” an Analytic Magazine article written by Irv Listig, Brenda Dietrich, Christer Johnson and Christoper Dziekan recently explained prescriptive analytics in greater depth.

Prescriptive analytics, which is part of “advanced analytics,” is based on the concept of optimization, which can be divided into two areas:

  • Optimization: How can we achieve the best outcome?
  • Stochastic optimization: How can we achieve the best outcome and address uncertainty in the data to make better decisions?

Once the past is understood and predictions can be made about what might happen in the future, it is then time to think about what the best response or action will be, given the limited resources of the enterprise. This is the area of prescriptive analytics. Many problems simply involve too many choices or alternatives for a human decision-maker to effectively consider, weigh and trade off – scheduling or work planning problems, for example. Twenty, 15 or 10 years ago these problems could only be solved using computers running algorithms on a particular data set for hours or even days. It was not useful to embed such problem-solving capability into a decision support system since it could not provide timely results. Now, however, with improvements in the speed and memory size of computers, as well as the significant progress in the performance of the underlying mathematical algorithms, similar computations can be performed in minutes. While this kind of information might have been used in the past to shape policy and offer guidance on action in a class of situations, assessments can now be completed in real time to support decisions to modify actions, assign resources and so on.

Behind all the big words and new terms the idea is simple. Prescriptive analytics is a tool for making choices that will create real benefits for your organization.

Of course, you can never get to this level if you don’t start with the proper training. Why not register for the December 2013 webinar and learn how to properly launch your own analytics projects today?

UK Supermarket Chain Uses Big Data to Reduce Energy Costs

TescoThe UK supermarket giant Tesco already uses big data in all of the ways that are standard to the industry. They use loyalty card data and data on the ways that customers interact with the store to drive more sales and create good promotions. Now, however, they’re using big data to tackle a new problem–energy consumption and costs in their stores.

The initiative targets in-store refrigerators and other equipment. Computer Weekly explained the project in a recent article.

“The move will help the retailer cut its refrigeration energy costs by up to 20% across 3000 stores in the UK and Ireland.

…The project, which used sophisticated computer systems to analyze gigabytes of refrigeration data, revealed that, without realizing it, many Tesco stores in Ireland were running their refrigerators at a lower temperature than necessary.

“Ideally, we keep our refrigerators at between -21 degrees Celsius and -23 degrees Celsius, but in reality we found we were keeping them colder. That came as a surprise to us,” said John Walsh, Tesco’s energy and carbon manager for Ireland.

…The data warehouse takes readings every three seconds from in-store sensors, processes the data in real time, and displays the results on a Google map which shows the performance of refrigerators in more than 120 Irish stores. It is also able to monitor and control the performance of Tesco’s heating and lighting systems.”

Whether it’s making changes, cutting costs, detecting fraud or predicting the effects of proposed changes in a store, big data, properly utilized, is a big help to business.

It all starts with getting the proper training on how to make the best possible use of that data. The Modeling Agency offers a free webinar to help business owners get a handle on the possibilities, and TMA’s other predictive analytics training courses can help you refine and strengthen your strategy so that your business enjoys the best possible results for its efforts.

Research Report Reveals How Top Companies Use Big Data

BoardroomThe Silicone Angle blog recently covered some of the insights to be found in an SAS and International Institute for Analytics report called “Big Data in Big Companies.” The report is available as a free download on the SAS website.

Each of the companies that the report covered spoke of using data to make smarter business decisions. Big data, when it’s functioning properly, creates real value for the companies who choose to make use of it.

One case study came from UPS. It’s a good study, one that comes with impressive numbers.

“UPS has more experience with Big Data than most, given that it first began tracking the movements of its vehicles and the packages it delivers back in the early 1980s. The firm reckons it tracks in the region of 16.3 million packages a day, whilst dealing with 39.5 million tracking requests from its customers each day. To date, the company has now accumulated over 16 petabytes of Big Data.

UPS’s ORION (On-Road Integrated Optimization and Navigation) initiative is said to be one of the largest operations research projects in the world. The majority of its data comes from telematics sensors installed in its vehicles, together with mapping data and other real-time reports of drop offs and pick-ups from drivers.

According to UPS, the ORION initiative has helped it to shave around 85 million miles off its daily routes during 2011, something that saved it more than 8.4 million gallons of fuel. The company estimates that this has led to $30 million in savings.

GE provided another great example. They use data to prevent mechanical failure.

“Bill Ruh, Vice President and Corporate Officer for GE’s Global Software Center…highlights GE’s industrial business as a prime target for big data, referencing the health of blades on the jet engines the company manufactures. “Our sensors collect signals on the health of blades on a gas turbine engine to show things like ‘stress cracks.’ The blade monitor can generate 500 gigabytes per day–and that’s only one sensor on the turbine. There are 12000 gas turbines in our fleet.” The value in integrating all the sensor data onto a big data platform can reveal patterns on when blades break, allowing GE to time its manufacturing and repair process before a break occurs.

The big companies all use data in ways that are as unique as their businesses are. However, they each have something in common, as well. Each company has set sensible targets for their business which allows them to create real, actionable results.

This kind of effectiveness starts with the proper training about data analytics. If you’re ready to put data to work for your business, start with TMA’s free webinar. Then, take a look at any of The Modeling Agency’s predictive analytics training courses to get more insight about what big data can do to help your business.