Making Statistics Work in the Real World
Ask about Knowledge at Wharton
Episode Description
Recent Wharton research examines the interplay between the computational complexity of statistical methods and their performance. More: http://knlg.net/2iFryNx 📂 Management;Technology 🏷️ Research
Read the full transcriptHide the full transcript
- we're here today with Wharton statistics Professor bashor bataria to talk about some of his latest research bashor
- thanks for being with us today could you first of all talk to us a little bit about give us a brief summary of your
- research what kind of question you were trying to answer right so my research interests are the intersection of
- Statistics probability and combinatorics so recently numerous very interesting combinatorial and graph theoretic
- problems of emerging statistics mainly because of the ubiquitous presence of network data and the increasing use of
- graph based methods in modern analytics and as a consequence many interesting connections have emerged between modern
- statistical methods and classical Concepts in geometry and probability and you know as a consequence you can use
- them to solve interesting problems in statistics great and tell us a little bit about in this study what were the
- key takeaways that you took away that you took away from the study the key takeaways of my research basically the
- interplay between computational complexity which is the time it takes to to implement one of these methods and
- its statistical performance perance which is how close it is to the mathematically based procedure it turns
- out that many of these graph based methods are computationally very efficient so they can be aply used to
- solve uh and apply in large data sets and we have also shown that in many situations these tests have near Optimal
- Performance guarantees which provide the theoretical justification required for using these procedures now had graph
- based methods is that something that has not been used as much in the past or had
- been looked at with more skepticism mean why is this right so so graph based methods have
- been used in practice for a long time but I think what comes out of my research is the the answer to the
- question that why they work so they were used and they were working fine before but here is so here we provide some
- theory behind why it works giving the justification of using it in real problems great and so getting to that
- actually like if I'm a business if I'm a business person or I own a business I mean how can I apply This research
- practically in my life right so one of the recent projects we're looking at is what is known as the two sample problem
- so imagine I have a situation where I want to find out whether a set of genes regulate or affect the occurrence of a
- disease so for example to just illustrate suppose I have uh uh 20 genes and I have the gene expression level
- data from 100 patients who have diabetes and I have the same expression level data for 100 patients who are healthy
- and the goal is to find out whether these 20 genes are expressed differentially so by that I mean that
- their expression levels are sign signicantly different between these two sets of patients and uh our research
- sort of aims to provide the theoretical understanding and comparison of the different methods that are deployed to
- understand or answer such questions what are some other applications for This research right so another interesting
- application of our work is in natural language processing mainly in problems which try to understand similarity
- between words so imagine the word color which can be spelled in two ways one with the letter U one without the letter
- U and they are the same words but the words for instance wolf and fox are both animals but they are very different
- words so in this case in spite of the amount of data we have the support size which is basically the collection of all
- words is far larger than the than the data set itself so one of our methods is what we are studying can be used to
- analyze such problems as well which I would think would be pretty interesting to businesses nowaday just because so
- many people are like taking social media posts and things like that trying to get
- data about customers that way right perfectly yeah yeah yeah sure and so what's next for This research right so
- currently I'm trying to understand or or analyze the methods for analyzing data in what is known as the high dimensional
- setting where you have say 10,000 genes and only a few hundred patients and uh you want to find out something about how
- the genes affect the disease or the patients and for these cases different new techniques are required and I'm
- trying to understand the theoretical uh the uh background of these results and how these can be used to find to get new
- methods and new algorithms great BR thanks for being with us today thank you thank you
- [Music]
Brief Summary
Wharton Professor Bashor Bataria discusses graph-based methods in statistics and their applications in genetics and natural language processing.
Key points
- Exploring Graph-Based Methods. Professor Bashor Bataria discusses the efficiency of graph-based methods in statistics.
- The Two Sample Problem. A practical application of the research involves analyzing gene expression levels.
- Applications in Natural Language Processing. The research extends to understanding word similarities in language processing.
Episode highlights
- Exploring Graph-Based MethodsProfessor Bashor Bataria discusses the efficiency of graph-based methods in statistics.“Many of these graph-based methods are computationally very efficient.”1:05Jun 29, 2017
- The Two Sample ProblemA practical application of the research involves analyzing gene expression levels.“The goal is to find out whether these 20 genes are expressed differentially.”2:06Jun 29, 2017
- Applications in Natural Language ProcessingThe research extends to understanding word similarities in language processing.“Our methods can analyze problems in natural language processing.”2:46Jun 29, 2017



