Slashdot Mirror

← Back to Stories (view on slashdot.org)

DePaul University To Offer Degree In Predictive Analysis

Posted by samzenpus on from the tea-leaves-and-chicken-bones dept.

itwbennett writes "The Chicago-based DePaul University will offer what it says is the nation's first master's degree in predictive analysis, the school announced on Wednesday in conjunction with IBM, which will provide resources for the program. 'We realized there was a need to create a program that prepared students in careers in data analytics and business intelligence,' said Raffaella Settimi, an associate professor at DePaul's College of Computing and Digital Media, who helped craft the program. 'A lot of the professionals who work in these fields have a variety of backgrounds, but there really isn't a program dedicated to data analytics,' Settimi said."

1 of 71 comments (clear)

  1. Went to DePaul for Statistics by proc_tarry · · Score: 4, Informative

    I went to DePaul and got an MS in Applied Statistics. I also work for a marketing company doing "predictive analytics". (I really don't like that term. It degrades the importance of understanding the relationship between independent and dependent variables, and places more emphasis on dependent variable fit. You can have a highly accurate model, but if marketers don't understand how their efforts affect sales the model is worthless.)

    At DePaul one series of classes was mostly math theory, the remaining classes were 100% about "predictive analytics", i.e. using a computer to build statistical models. It used a more traditional approach to applied statistics with "topical" classes: sampling, forecasting, design & analysis of experiments, nonparametric statistics, Monte Carlo simulation, multivariate statistics, etc.

    The statistics program is part of the College of Liberal Arts & Sciences in the math department. This new program is in the computer science department (which has its own college). The program (http://www.cdm.depaul.edu/academics/Pages/MSinPredictiveAnalytics.aspx) looks like a hybrid of CS, Stats & Marketing. It includes a class in neural networks which most stats programs lack, but again, this focuses on "prediction" instead of "inference", which is less useful to marketers. (Neural networks is a highly valuable topic. Just not as much in this field.) Also, the program lacks pure programming classes, which there is A LOT of in this field (data never comes in formats ready to model). Most is done in SAS or R, but any programming language teaching basic concepts (variables, logic, arrays, loops, functions) is useful.