Big Data & Analytics Innovation Summit, Australia: Day 2 Highlights
Tags: Analytics, Big Data, Conference, Consumer Insights, IE Group, Smartphone, Sydney-Australia, Text Mining
Highlights from the presentations by Big Data leaders from Paypal, Huawei and Qantas on day 2 of Big Data & Analytics Innovation Summit 2014 in Sydney, Australia.
The Big Data Innovation & Analytics Summit (Sept 17-18, 2014) was organized by the Innovation Enterprise in Sydney, Australia. It brought together a large gathering of senior business executives leading Big Data initiatives in Australia. The summit brought together business leaders and innovators offering solutions and insight in the Big Data space. Big Data Innovation helps utilizing data-driven strategies and discovering disciplines that change because of the advent of data. With a vast amount of data now available, modern businesses are faced with the challenge of storage, management, analysis, visualization, security and disruptive tools & technologies.
Highlights from Day 1.
Here are highlights from Day 2 (Sept 18, 2014):
Daniel Nelson, Braintree Head of Enterprise Development, PayPal offered valuable insights in his talk "Smart Phones & Big Data". The future is here, it's just not evenly distributed yet. Daniel Nelson described how the mobile revolution is one of the most significant frontiers for big data and offers both massive opportunity and significant hazards for brands. Commerce is undergoing enormous disruption caused by the exponential growth of smart phones. Smart Phones have already taken over online commerce and the long awaited offline takeover is building momentum.
72% of Australians said they never leave home without their smartphone. They check their smartphones an average of 150 times per day. 60% of total time on social media is spent on smartphones. Smart phones come bristling with sensors, connectivity and their owner’s unrivaled attention, making them a Data Scientist’s dream. By 2017, there will be over 20 billion mobile connected devices. The smartphone will be the interface to the "Internet of Things (IOT)". Smartphones and Cloud are symbiotic trends. Smartphones are the collection and delivery point for Big Data. Big Data is redefining e-commerce in many significant ways - analytics to optimize traffic & conversions, dynamic & optimized pricing, recommendation systems, pay-for-performance advertising, attribution modelling, etc.
He ended his talk stating that smartphones enable Big Data strategies to break free from online and to flourish and grow in a mobile-enabled world.
John Vidler, Director, Solutions Architecture, Huawei gave an interesting talk on "Making the Big Data Connection". Telecom industry is facing the challenge of addressing the seemingly endless increase in data collection and the expectations on businesses to manage it in a meaningful way. The Big Data challenge threatens to increase IT budgets and staff requirements by an estimated 40% and potentially cause organisational disruption instead of graceful evolution.
Citing Big Data and 4G as the major telecom trends, he mentioned that several lessons from the 4G roll-out apply to Big Data, such as the importance of speed and user experience, data-centric packages, the transforming profile of data usage and increasing trend of digitization. Big Data, a new productive force within society, can generate significant financial value across sectors. However, value discovery from Big Data is the biggest challenge en-route. Today, Big Data is triggering wide-scale commercial reform - disruptive products and services are emerging, new business models are launching at an unprecedented pace, etc. From telecom industry perspective, customer experience is driving Big Data. Most customers associate Big Data with real-time information.
Through real-life case studies, he explained how Big Data can provide great business insights such as consumer profiling, deeper understanding of purchase decisions, etc. For Business-to-Consumer (B2C) segment, business value can be reaped through incremental offers (in terms of bandwidth and download limit) in order to grow user consumption. In the Big Data era, user centrality and awareness drive value growth. Thus, it is important for companies to gain insights into user behavior analysis through harnessing Big Data.
He described an ICT platform to drive decision making into real-time operations; along with B2C as well as B2B use case scenarios. In conclusion, he stated that Big Data brings new challenges as well as new opportunities; and an appropriate strategy will enable monetization.
Con Menictas, Senior Statistician, Qantas delivered an insightful talk on "Text Mining Offers the Best of Both Worlds". Text Mining (TM) is being talked about more and more in the marketing research space, but how does it compare to traditional research approaches such as qualitative and quantitative research? Qualitative research can exact great richness of insights but is not generally projectable to the population. Quantitative research is generally projectable to the population but lacks the richness of qualitative research. However, text mining analytics contain the richness of qualitative research and the capability of projecting to the population!
Currently there are many approaches towards text mining - word clouds via content, sentiment towards brand, thematic representations, text time-series analysis, concepts via clustering, etc. Very often, the key problem is to build predictive models to predict satisfaction, recommendation or NPS (Net Promoter Score).
He worked on this problem using text data only, specifically from just one question posed to the subjects, to demonstrate that a single verbatim open ended question can provide predictors of satisfaction. He presented a case study, which compared qualitative and quantitative research outputs, to those obtained via mining and modeling of verbatim text data. Through the case study, he demonstrated that TM provides advantages in cost; time; resources; and scalability, when compared to more traditional methods such as structured qualitative and quantitative research. TM produced key client specific drivers, in addition to insights on how to implement those drivers.
Explaining the mechanics of text mining, he said that before applying analytics the text data needs a lot of pre-processing - stemming, punctuation removal, number removal, applying dictionaries, applying thesaurus, case changes, stop word removal, etc. He stressed that the text mining approach must be faithful to each body of text because of the extreme distributional demands of concept and theme generation.
The results of his approach showed that text derived variables contain latent structure closer to reality than loose scale construction that is commonly used in the industry. He also noted the importance of open ended questions which do not contain respondents to the researcher's framework, and are a lot easier than other many types of scales.
He believes that TM offers the marketing research industry another arrow in its bow for two reasons. First, text mining offers insights that would otherwise go unnoticed because it is based on free-form text. Second, TM is fast, low-cost, and can have results to clients within days as opposed to weeks.
Related:
