Big Data & Analytics for Retail Summit 2014 Chicago: Day 1 Highlights
Highlights from the presentations by Big Data leaders from Sony Pictures Entertainment, Macy's and Nuevora on day 1 of Big Data & Analytics for Retail Summit 2014 in Chicago.

The Big Data & Analytics for Retail Summit (June 19 & 20, 2014) was organized by the Innovation Enterprise at Chicago. Illustrated intermittently with case studies, interactive panel sessions and deep-dive discussions, this summit offered solutions and insight from the leaders operating in the Big Data space.
We provide here a summary of selected talks along with the key takeaways.
Here are highlights from Day 1 (Thursday, June 19, 2014):

A traumatized economy and the proliferation of disruptive business models has driven simultaneous transaction growth & margin decline and the erosion of 2 key profit drivers: (1) new release "conversion rates" and (2) consumers' purchase of "catalog" titles. As a result, consumer retail is getting more front-loaded, i.e. they have just one chance to get it right. His firm had to tighten their "greenlight" models, as prior models failed to forecast about 30% of profit opportunity.
He shared a case study on: when does an Oscar boost sales, to what degree and why? He also mentioned that Analytics has helped the firm change how it thought about its retail partners.

Big Data is a big opportunity, thanks to Moore's law working on computing, storage, and networking capacities. Data Analytics has changed every field significantly from science to government to commerce. In the traditional BI process, data can be accessed and analyzed only after ETL (Extract-Transform-Load) process. In terms of data maturity, most companies are still at best doing segmentation and predictive modeling or creating multi-dimensional reports; and thus, haven't reached the stage of knowledge discovery.
In the Big Data era, data modeling faces the following challenges:
- Modeling needs to scale
Today's data models are using thousands of predictors, compared to just a few being used in traditional models. Data modeling has to be automated, as human interactive model building is not scalable. -
Timeliness of models
As data is getting cheaper, data models may get outdated in weeks. If we cannot build and update models in time, we cannot benefit from many useful patterns. -
Time and effort required for data integration
Measurement and attribution are key. Analytics teams need to think in terms of model metrics, and accumulate assets of creative, best practices. -
Test and experimentation
Customer response behavior is complex. Organizations need to test the different product versions, new models and new messages. They must split traffic tests for web or email. It is important to identify test design problems and understand their implications.
He outlined the solutions for above challenges as:
- Big Data warehouse solutions
The terabytes of data we need to handle can take hours just to scan it. Ideal solution requires a cloud of servers with local storage with capabilities to read, process and write intermediate results in parallel. -
Separation of concerns
Data Science needs to address a wide variety of concerns, such as solution complexity, data complexity, requirements variability, reliability, scalability, and latency. -
Scalable modeling tools
The need is to build robust models that can handle missing variables and outliers. Model optimization should be automated. Models that are unnecessarily large must be penalized. -
Follow the best practices in modeling
Analysts must understand how the data are collected: what data can and cannot be collected. They need to balance cost of collecting data and optimize modeling. Remember that good ideas are not necessarily complicated. The focus should be on domain knowledge, and not just on the data mining tools.
Finally, he talked about the Advanced Analytics team at Macy's. The team uses a wide range of tools including Hadoop, SAS, SAP/KXEN, R and Mahout.

Though it might seem deceptively simple, this Analytics process has a number of challenges. One of the biggest challenges is the disparate and large number of data sources. Referring to a study from Aberdeen (2012), he mentioned that Big Data initiatives are delivering tangible business benefits. Firms leveraging Big Data have 3% more profit, 10% greater operating cash flow and 4% more customers year-over-year compared to firms not leveraging Big Data.
It's important to be able to plunge into Big Data, but even more so to know how to pick the RIGHT data and drive your focused objective. The "Right" Data trumps "Big" Data any day. He suggested that companies should think "Process" and act "Value Driven" They must optimize value drivers in the context of their business process, while remembering that optimizing a business process is an asymptotic process.
While there are a lot of Big Data apps that can be used for various processes across the Customer Marketing lifecycle, it is very important to focus on the most important processes based on business context. He gave a high-level overview of the Nuevora platform which delivers integrated Analytics apps alongwithcontinuous monitoring. In the end, he emphasized that "Context is Key" and the importance of making Analytics easy to use - "Consumerization".
Highlights from day 2.
Related
Top Stories Past 30 Days
|
|