The Fifth Elephant 2017

On data engineering and application of ML in diverse domains

Causal Analytics in Retail and Telco

Submitted by Gaurav Goswami (@gauravgoswami) on Friday, 28 April 2017

videocam_off

Technical level

Intermediate

Section

Crisp talk for data engineering track

Status

Submitted

Vote on this proposal

Login to vote

Total votes:  +2

Abstract

In this talk, I will discuss causal analytics using machine learning in the retail and telco domains. This talk should provide a brief overview of the value machine learning can provide in these domains along with the associated challenges and opportunities.

Outline

The key aspects of the talk are:

  • Introduction to the need for causal analytics in retail and telco settings: identification and explanation of anomalies
  • Discussion on the challenges and opportunities
  • Using external data for additional signals in the retail setting
  • Challenges in automating causal relationships only using observational data
  • Overview of possible approaches and expected outcomes

Speaker bio

I am currently working at IBM India as an Artificial Intelligence/Machine Learning expert. My Ph.D. thesis focuses on using machine learning and computer vision techniques to solve challenges in face recognition by improving the computation and combination of robust representations. I have had the opportunity to be a part of crafting a machine learning based solution for real world use cases in these domains. This talk will provide a brief overview of my experience and perspective on some of the key aspects of applied causal analytics.

Links

Slides

https://www.slideshare.net/secret/pCloJCaYMkhnMa

Comments

  • 1
    Nina T (@ninat) a year ago

    There are no details given in the slides. Could you please update them?

    • 1
      Gaurav Goswami (@gauravgoswami) Proposer a year ago

      Yes, I just posted the outline thus far. I will post detailed slides in a day or two.

      • 1
        Zainab Bawa (@zainabbawa) Reviewer a year ago

        The slides barely provide any information. Detail out the content in the slides, explaining the full flow of content, and also upload a preview video explaining what your talk is about and key takeaway for the participants.

        • 1
          Gaurav Goswami (@gauravgoswami) Proposer a year ago

          Thank you for the feedback. The initial slides were just the outline. I have uploaded the full set of slides now which should reflect in the existing slideshare link in around half an hour from now. I will upload a preview video before end of day today as well.

  • 1
    Gaurav Goswami (@gauravgoswami) Proposer a year ago

    Video preview/summary is now available at: https://youtu.be/Hq6ll9_NaXk

  • 1
    Zainab Bawa (@zainabbawa) Reviewer a year ago

    You have mentioned use of Watson. Is this open source and/or can these problems be solved with tools other than Watson which are open source?

    • 1
      Gaurav Goswami (@gauravgoswami) Proposer a year ago

      Yes, there are open source tools/libraries that can also be used to achieve similar functionality.

      • 1
        Zainab Bawa (@zainabbawa) Reviewer a year ago

        Will you be covering these? Because Watson is not open source. If the talk only focusses on Watson, then we cannot consider it under the open submission process. It becomes a talk on proprietary tool where there is no takeaway for the audience.

        • 1
          Gaurav Goswami (@gauravgoswami) Proposer a year ago

          Yes, the majority of the talk will not be specific to any one platform and will cover open source tools that exist to solve parts of the problem. Watson Discovery is a streamlined tool that can accomplish some of these tasks and is just 1 slide out of 12 in the talk. It essentially has certain NLP capabilities which are provided by various open source offerings in various programming languages with varying degrees of performance and complexity. It also offers an indefinite free trial period which is only limited in the volume of data, so the audience can explore it after the talk if they wish.

          • 1
            Zainab Bawa (@zainabbawa) Reviewer a year ago

            Thanks for this update, Gaurav. The proposal currently tries to cover too many topics. You can either pick anomaly detection or causal analytics. Also, what is the unstructured data you mention in the proposal? We’ll need to see more detailed slides, explaining how this talk is about the concepts and application, and that IBM Watson is incidental to the overall discusson.

Login with Twitter or Google to leave a comment