Join Wayne Winston for an in-depth discussion in this video What do you need to know to be an analytics professional?, part of Wayne Winston on Analytics.
A key road block in a lot of analytics projects is the fact that different data sources don't talk to each other. Let me give you an example. Suppose you work for a fast food restaurant. And you want to know, what factors determine the success of employees in your business. So you have one data base that measures how successful employees were in your business. And whether they left the business. High turn over rate is important to fast food business. You want to lessen that. And so, that's one database. So then another database would have basically what did you know about the employee when you hired him or her? The problem is that many companies these databases don't communicate.
So in other words, I would have the performance book of the employee in one database. I would have what I knew about the employee when I hired him or her in another database and they don't talk to each other. So it would be impossible to do any analysis to figure out what caused the employee to perform well or perform poorly. So you want those data sources to talk to each other. That's really an important roadblock resolving that before you can have success with analytics. Without the, all the data sources at the organization talking to each other successful analytics is almost impossible.
So then if you have the data sources talking to each other, how can you analyze it for prescriptive and predictive analytics. Well, you need to know some tool to do statistical analysis. So, with small data sets Excel will do a fine job. But, if you have really big data set, there are lot of other alternatives out there. So I think the analytics professional needs to be fluent in either SAS which is used at a lot of major corporations. R, which is predominant in the academic world or SPSS, which is now owned by IBM.
But you probably want to know something about all those packages. And lynda.com has training in some of these and, basically, you've got to know how to take that big data and get some meaning out of it. And it's not just enough to know how to basically write a SAS program, an R program. Or run a procedure in SPSS. You've gotta know something about statistics. You've gotta know something like Nate Silver knows a lot of statistics. He developed his algorithms using statistics. You've got to basically know a lot of statistics to be able to under separate the weak from the chaff and really understand what's going on.
And then two final areas the analytic professional should be proficient in, are optimization and simulation. So what is optimization? Finding the best way to do something. So we mentioned Eli Lilly. How can you optimize the yield of production process what settings would do that? Okay, we mentioned portfolio optimization, what asset allocation would be the best. Best given your risk tolerance. Given the desired expected return you want. So that's optimization, finding the best way to do something given constraint. And then final tool that I would like to mention is Monte Carlo simulation.
Modeling uncertainty. Like a key decision would be should you rent your house or you buy your house, okay? Well the answer depends on where you think the price of the house will go in ten years or whenever you sell the house. Or maybe you won't even be able to sell the house. You want to model the uncertainty in either. So, any situation in which you want to model uncertainty, you would use Monte Carlo simulation. Another example? Capital budgeting. Should Proctor and Gamble put out a new cosmetic? And, so basically, they don't know if it's going to make money, so they want to assess the odds Will it make money? If there's an 80% chance, lets say, a new cosmetic for Proctor and Gamble will make money, then basically they should go ahead with that.
- What is analytics?
- Predictive vs. prescriptive analytics
- What do you need to know to become an analytic professional?
- Looking at examples of good and bad metrics
- How mobile devices will shape analytics