I'll be offering a 4-week online course on Acceptance Sampling for Quality Control, starting April 20, 2012. In this age of outsourcing and off-shoring, sampling inspection is a necessary approach for assuring quality in the presence of many remote suppliers and vendors. Sampling inspection (known as "acceptance sampling") is used by many companies worldwide and is often based on ISO procedures. The ISO requirements are based on statistical and probabilistic principles. In this course, you will learn about different sampling inspection procedures, their logic, and how to implement them. The course will use my text Practical Acceptance Sampling: A Hands-On Guide [2nd Edition]
News
Online course "Acceptance Sampling" opens this Friday (Apr 20, 2012)
Keynote "De-Mystifying Predictive Analytics" @ ReTechCon (Mumbai, May 17-18)
Next month, I'll be delivering the keynote address at the annual conference of the Retailers Association of India, called the Retail Technology Conclave (ReTechCon). Here is the synopsis for my talk on "De-mystifying Predictive Analytics": Business Analytics, Big Data, and Predictive Analytics are the latest buzzwords promising to turn straw into gold for retailers. The media and software vendors have created an excitement about these technologies and the transformations that they can achieve. But what exactly are these technologies? How are they different from BI reporting or from statistical analysis? Without understanding the basics of predictive analytics, it is impossible to identify opportunities, to critically assess possibilities, and to figure out needed resources. Does "Big Data" require data on millions of customers for extracting non-trivial and potentially actionable knowledge? Can highly sophisticated algorithms replace domain knowledge? What is the value of connecting data from different sources in the business? This talk will de-mystify the basics of predictive analytics: the toolkit, data requirements, performance evaluation, the types of questions it can potentially answer, as well as the "fine print" of hidden assumptions. We will discuss opportunities of leveraging predictive analytics in the Indian retail market, including emerging online and mobile commerce.
"Testing Theories with Big Data: A Super-Power Approach" @ SCECR 2012
In the upcoming Statistical Challenges in ECommerce Research symposium I'll present a talk about the use of standard statistical modeling when using large samples. Many of the standard methods don't "scale up", and even when they do, there are many missed opportunities if a "big data" approach is not taken.
Effective Data Presentation Workshop @ Royal Institute of Health Sciences
Presenting data effectively is crucial for decision making. In public health, data arrive from multiple geographical locations and from multiple sources (labs, hospitals, health facilities, etc.). The ability to present such data by using effective charts is needed for creating reports and presentations that highlight the information.
I instructed a half-day hands-on session on Effective Data Presentation, as part of a week-long workshop on public health surveillance, organized by the Australian AID at the Royal Institute of Health Sciences in Thimphu. Participants were the first batch of newly graduated public health students.
Slides and further information can be downloaded from here.
Simulating multivariate count data - paper in print
Lots of studies are based on simulating count data from a Poisson distribution. Many simulate multiple Poisson variables that should be correlated, but are simplistically generated as independent. Our just-published paper "On Generating Multivariate Poisson Data in Management Science Applications" co-authored with Inbal Yahav introduces a practical method for simulating correlated multivariate Poisson count data -- highly useful in many applications! Here's the abstract (full paper here)
Generating multivariate Poisson random variables is essential in many applications, such as multi echelon supply chain systems, multi-item/multi-period pricing models, accident monitoring systems, etc. Current simulation methods suffer from limitations ranging from computational complexity to restrictions on the structure of the correlation matrix, and therefore are rarely used in management science. Instead, multivariate Poisson data are commonly approximated by either univariate Poisson or multivariate Normal data. However, these approximations are often not adequate in practice. In this paper, we propose a conceptually appealing correction for NORTA (NORmal To Anything) for generating multivariate Poisson data with a flexible correlation structure and rates. NORTA is based on simulating data from a multivariate Normal distribution and converting it into an arbitrary continuous distribution with a specific correlation matrix.We show that our method is both highly accurate and computationally efficient. We also show the managerial advantages of generating multivariate Poisson data over univariate Poisson or multivariate Normal data.