Application of Data Analytics in Industry - liveanalytics.org

Last week, we learned about the "SCIENCE OF ANALYTICS". This week we will talk about some of the successful applications of analytics. We will start this week with 2 different examples which I was reading recently from a podcast of Accenture and then move to some of the other industrial examples.

1. Harrah's Entertainment

Gary Loveman, CEO of Harrah's, was presenting at an event and sitting in the front row were all of his competitors. He stood up, he showed slide after slide that showed precisely how they use their loyalty card, how they did the analysis, what kinds of metrics they looked at. And their competition was writing everything down.

Finally, after about an hour of this, someone raised their hand and said, "Dr. Loveman, doesn't it bother you that your competitors are taking down every word you said?" He just looked quizzically at them for a minute and said, "No, not really, because by the time they figure out how to do what I just described to you, we will be so far ahead of them that they will never be able to catch up."

2. A. C. Milan: A professional football club based in Milan, Lombardy and one of the most successful organizations. \

• Background: A.C. Milan had a situation that was very high-profile and very embarrassing. They hired a player for a lot of money, very expensive. Within a couple of weeks, he promptly blew out his knee and was absolutely useless to them for the rest of the season.
• Objective: They realized that they needed to look at not just what somebody's previous scoring capability was, they needed to look at their potential for playing and contributing to the team. That came down to keeping them healthy.
• Approach: So, they actually formed a research group that analyzed every aspect of a player's moves-how they run, how they jump-and analyzed the likelihood that they were going to become injured.
• Outcome / Impact: Using that data, they used it initially to decide who to purchase and who not to purchase. But, over time, they started using it in a different kind of way. They were able to work with the players to help them understand, "You're turning your left foot out too much. You are going to injure your ankle if you do that." They actually now meet with each player about twice a month to help them analyze the latest data around their movements. They actually analyze an incredible amount of data; 50,000 data points about a single player; 200 just about their jump. This is a company that is really trying to take analytics to the next level.

3. Financial Industry (Credit Card) - Marketing Analytics

• Background: A credit card company has a marketing budget of Rs. 1 cr or 10 lac pieces of mail set aside for sending out direct mails. If they send out mails to all the available lists that they have, they would need to spend Rs. 10 cr or 1 cr pieces of mail. They also know that by running this marketing campaign they will receive at the most 50 thousand new customers.
• Objective: From the 1 cr available pieces they would like to identify the 10 lac which are most likely to respond to the offer so as to increase the company's customer base and in turn profitability.
• Approach: Review data from their past 5 direct mail campaigns and build a predictive model which can help them identify likelihood to respond and differentiate "High probability to respond prospects" vs "Low probability to respond prospects".
• Outcome / Impact: Using the data from past campaigns, they were able to build a logistic regression (statistical predictive) model. This model looked at the history and helped them identify the right 1 cr to mail to book 40 thousand accounts. What it means is that for the rest of 9 cr, they would have booked only 10 thousand accounts (extremely inefficient). This exercise is carried out in majority of direct marketing to ensure the money spent has optimal impact. By leveraging the history this company was able to book 80% of accounts with only 10% of the mail and hence really reducing their cost to book the accounts.

4. Progressive Insurance Industry (Motorcycle Insurance) - Pricing & Risk Analytics:

• Background: A few years ago everyone was treating motorcyclists the same as if they had the highest risk and needed the highest price for insurance. They were not good credit risk. Everybody knew that and it was conventional wisdom.
• Objective: Identify segments within the motorcycle insurance owners who have lower credit risk than average.
• Approach: Review data from their past and "DE AVERAGE THE RISK". They used historic data to determine pockets / segments of customers with "HIGHER THAN AVERAGE RISK" and "LOWER THAN AVERAGE RISK".
• Outcome / Impact: What they found is that while some motorcyclists are really high risk, a majority of them are not. Eg a teacher driving a motorcycle is a lot lower risk than a high school student driving a motorcycle. With this in mind, they were able to offer much lower price to the teacher which increased their receptiveness to the low risk segment and thus increasing their customer base. Progressive has made a real art of skimming these sub-segments / pockets off, carving them out, focusing on them, serving them very well and moving on before anyone notices what has happened to them. And that's really the heart of the strategy for them.

That is really what makes analytics a sustainable differentiating strategy for these companies. It is because it is not about a single insight; it is about a set of processes they have, a way of using data and incorporating it into their decision making, that really helps them transform their business. It makes them much more able to maneuver changing business conditions; it makes them much more likely to anticipate changes in customers and markets; and, most importantly, it allows them to come up with different scenarios and understand how they ought to react to these changing market conditions.

Website: http://www.liveanalytics.org

Control Healthcare Costs Using Data Analytics - liveanalytics.org

When you look at your budget, you are likely going to see a significant amount of money being dedicated towards healthcare costs. Even if you are not self-funding your health insurance program being offered to employees, you are still contributing something - and this may be getting out of hand. The best overall solution is to control healthcare costs using data analytics.

There are a number of companies that are beginning to offer healthcare analytics in terms of a benefit optimization system.

What is a benefit optimization system? Depending upon the company you work with, the optimization system is going to warehouse data, clean it up, and provide the data to you in a visual dashboard where you will be able to customize queries of the data. As you are able to form queries, you will be able to obtain a significant amount of information as it pertains to your healthcare costs. You will be able to control healthcare costs when you know how much you are spending, what your return on investment is, what the quality of life of your employees are, and how many people are being affected.

The reason you may be spending too much money on health care right now is because you do not have access to the right amount of data. If you do not know what the general well-being of your employees are, you are likely spending too much money. You have to find out what the risk factors are of your employees and how they are managing chronic diseases. More and more employees are suffering from such things as obesity, diabetes, heart disease, and various other health factors. You will only be able to control healthcare costs if you know about these things so that you know how to make improvements upon them with disease management.

It is already assumed that you are going to spend some money on your employees outside of health insurance. This may be employee wellness programs, fitness centers, prepared food, and various other items. You may be able to enhance employee well-being through prevention if you use healthcare analytics to learn more about the needs of your employees. You may be taking the wrong approach with your employees simply because you do not have the right data in front of you.

You will be able to control healthcare costs significantly when you use data analytics with the right benefit optimization system simply because you will be receiving recommendations. You will be receiving recommendations on actions you can take as well as ideas on how to increase return on investment an employee well-being. All of these recommendations are priceless in terms of reducing costs in determining whether you are offering the best healthcare plans to your employees.

With data analytics, you can use a variety of variables to process the data. You can learn about the average age of your employees, the geography, the gender ratio, and how risk factors and Occupational Health & Safety are affecting your employees. You simply need the right system in place to be able to control healthcare costs.

Website: http://www.liveanalytics.org

Data Mining - Efficient in Detecting and Solving the Fraud Cases - analyticsforprofit.com

Data mining can be considered to be the crucial process of dragging out accurate and probably useful details from the data. This application uses analytical as well as visualization technology in order to explore and represent content in a specific format, which is easily engulfed by a layman. It is widely used in a variety of profiling exercises, such as detection of fraud, scientific discovery, surveys and marketing research. Data management has applications in various monetary sectors, health sectors, bio-informatics, social network data research, business intelligence etc. This module is mainly used by corporate personals in order to understand the behavior of customers. With its help, they can analyze the purchasing pattern of clients and can thus expand their market strategy. Various financial institutions and banking sectors use this module in order to detect the credit card fraud cases, by recognizing the process involved in false transactions. Data management is correlated to expertise and talent plays a vital role in running such kind of function. This is the reason, why it is usually referred as craft rather than science.

The main role of data mining is to provide analytical mindset into the conduct of a particular company, determining the historical data. For this, unknown external events and fretful activities are also considered. On the imperious level, it is more complicated mainly for regulatory bodies for forecasting various activities in advance and taking necessary measures in preventing illegal events in future. Overall, data management can be defined as the process of extracting motifs from data. It is mainly used to unwrap motifs in data, but more often, it is carried out on samples of the content. And if the samples are not of good representation then the data mining procedure will be ineffective. It is unable to discover designs, if they are present in the larger part of data. However, verification and validation of information can be carried out with the help of such kind of module.

Website: http://www.analyticsforprofit.com

What You Need to Know About Popular Software - Data Mining Software - analyticsforprofit.com

Simply put, data mining is the process of extracting hidden patterns from the organization's database. Over the years it has become a more and more important tool for adding value to the company's databases. Applications include business, medicine, science and engineering, and combating terrorism. This technique actually involves two very different processes, knowledge discovery and prediction. Knowledge discovery provides users with explicit information that in a sense is sitting in the database but has not been exposed. Prediction is an attempt to read into the future.

Data mining relies on the use of real-world data. To understand how this technology works we need first to review some basic concepts. Data are any facts whether numeric or textual that can be processed by a computer. The categories include operational, non-operational, and metadata. Operational or transactional elements include accounting, cost, inventory, and sales facts and figures. Non-operational elements include forecasts and information describing competitors and the industry as a whole. Metadata describes the data itself; it is required to set up and run the databases.

Data mining commonly performs four interrelated tasks: association rule learning, classification, clustering, and regression. Let's examine each in turn. Association rule learning, also known as market basket analysis, searches for relationships between variables. A classic example is a supermarket determining which products customers buy together. Customers who buy onions and potatoes often buy beef. Classification arranges data into predefined groups. This technology can do so in a sophisticated manner. In a related technique known as clustering the groups are not predefined. Regression involves data modeling.

It has been alleged that data mining has been used both in the United States and elsewhere to combat terrorism. As always in such cases, those who know don't say, and those who say don't know. One may surmise that these anti-terrorist applications look for unusual patterns. Many credit card holders have been contacted when their spending patterns changed substantially.

Data mining has become an important feature in many customer relationship management applications. For example, this technology enables companies to focus their marketing efforts on likely customers rather than trying to sell to everyone out there. Human resources applications help companies recruit and manage employees. We have already mentioned market basket analysis. Strategic enterprise management applications help a company transform corporate targets and goals into operational decisions such as hiring and factory scheduling.

Given its great power, many people are concerned with the human rights and privacy issues around data mining. Sophisticated applications could work its way around privacy safeguards. As the technology becomes more widespread and less expensive, these issues may become more urgent. As data is summarized the wrong conclusions can be drawn. This problem not only affects human rights but also the company's bottom line.

Website: http://www.analyticsforprofit.com