Big Data for Managers

Peadar Coyle

  

@springcoil
 


My Experience


  • Expert in Big Data Technologies and Data Science
  • Mathematician by training - specialized in Machine Learning

  • Supply Chain Management
  • Customer Churn optimization for B2B in the Pharmaceutical industry
  • Customer Segmentation for Event Management at Letsface.cn
  • Sentiment analysis for a large financial consultancy
  • Web crawl data analysis at Import.io

I did Data in Shanghai

Helping an event website target its customers

Data in London

Data geeks need tools like these, I helped launch the beta.

I worked with Data in Luxembourg

For a small e-commerce website. Doing Supply Chain models.

Worked In Air Traffic Management

BTW Air travel produces A LOT of data!

Energy analysis is also interesting





Like analyzing quantitatively the price risk in our markets
 

So how come I care about data?


Well I always loved science.
I wanted to be a neuroscientist


Then I fell in love with Physics

I studied Quantum Mechanics and
Quantum Optics at Bristol

This gets a bit complicated...

And my cat was never much use...

So I ended up in Math & Stats...

Along the way I learned some programming and other skills...

I needed to find a career


And I decided Academia wasn't for me.


So I became a data scientist!


Now what skills does a data scientist have?




 To me data Science is a lot like Science


But the hardest thing to learn has been..


“Being a data scientist is not only about data crunching. It’s about understanding the business challenge, creating some valuable actionable insights to the data, and communicating their findings to the business.”

Jean-Paul Isson, Monster Worldwide, Inc.


I'm still learning the tech stuff too


What is Machine learning?
Well this next slide might help...


What is Data Science?

Data Scientists help you harness the value of 'big data'

So, who is talking about Big Data?


"We project a need for 1.5 million additional managers and analysts in the United States who can ask the right questions and consume the results of the analysis of Big Data effectively." -  Big data: The next frontier for innovation, competition, and productivity, McKinsey report


Who else?

Gartner says 'Data is the new oil'

'Data are becoming the new raw material of business'


But talk is cheap...

Big data is like teenage sex: everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it... -

Professor Dan Ariely - Duke University


So how do you get value out of your data?

You could hire a data scientist

This talk is aimed at helping you understand if you need to hire a 'data scientist'.

Aims of the talk

  • Explain the substance behind the phrase 'big data'



  • Tell you how you can use data in your business.



  • Help you understand the importance of data in your business strategy.

Or this talk could have been titled...

But Peadar where are the business examples?

 Example: Linkedin


But I work in the real world not online!

UPS uses data to travel more efficiently and
save millions on fuel consumption.

What is "Big Data" anyway?

Customer preference data is...

What is "Big Data" anyway?

Wind sensor data is too

  • 'Source: English Wikipedia, original upload 15 July 2004 by Leonard G. - Creative Commons Sharealike license

What is "Big Data" anyway?

Web crawling data is too

What is "Big Data" anyway?

Audio and visual data is too


What is "Big Data" anyway?

Social Media data must be too

What is "Big Data" anyway?

Not to mention social media metadata

 

What is "Big Data" anyway?




Genomics and health data are too.

What is "Big Data" anyway?



And what about Particle Physics data?

 


I hope you can see that there is...

Remember this slide?

So why good data analysis is hard?

  • Getting data is hard


  • Building models is hard


  • Asking the right business questions is even harder


I often have to borrow lots of peoples brains to get to
the right business questions...

 Pick the right methodology for the job


  • Text -> topic modelling, sentiment analysis, information extraction
  • E-commerce data -> prospensity analysis, collaborative filtering
  • Multimedia -> speech-to-text, audio fingerprinting, face recognition
  • Clickstream logs -> frequent pattern mining, sequence analysis
  • Proton-proton collision from LHC -> I have no idea despite having a Physics degree
And then what? Well you can tell stories with visualizations...
But Data Scientists don't just produce reports


They produce data products too.


So what is a data product?

Well I'm glad you asked...

What is a data product?


A data product provides actionable information without exposing decision makers to the underlying data or analytics.

Examples include: Movie Recommendations, Weather Forecasts,Stock Market Predictions, Production Process Improvements, Health Diagnosis, Flu Trend Predictions, Targeted Advertising.
– Mark Herman, et al., Field Guide to Data Science

    But you said...


    Here is an Example from Mailchimp:
    When should I send that email?


    Step 1 to 100: Data Scientists - do lots of analysis...

    And produce a magic button :)


    Define Data Science


    • It is the application of science and models to really complex human problems
    • Such as: Who is buying our product, why are they leaving our service.
    • Data scientists leverage mathematics and computer science to deliver business value such as smoother operations, enhancing your marketing strategy or  forecasting supply and demand.
    • In short data scientists help you prepare for the future.
    How does this help you in your business?

    Case Study - Marketing Analytics: In the Game Industry

    1) Uses gamers play data to optimize marketing
    communications across channels. - Customer segmentation modelling

    2) Building Personalization Engine Rules for 1:1
    communications with individual gamers. To help reduce customer churn.

    3) Predicts gamers likelihood to churn or to respond to
    up-sell offers.  

    Example

    Here is a graph of active users on an online game. Marketing teams use tools like this to monitor their customers in real-time

    What about forecasting? Do you mean the weather?

    (In Ireland and the UK it is quite easy - just guess rain all the time!)

    But there are other kinds of forecasts such as supply chain forecasts or demand forecasts....

    Example:Supply Chain Management

    • These are examples from Tableau an excellent data science product - based on laptop sized data sets. Similar to my Amazon work.  However these can also be built with open source tools.

    Electricity Demand prediction

    Electricity demand is well studied
    and in open source libraries.

    But predicting the future is hard


    "It’s Difficult to Make Predictions, Especially About the Future" -

    Niels Bohr


    So you can do an experiment


    And let data be your guide.

    User Conversion after a website change.


    Luckily they measured it. They learned the website change was a bad decision.

    Sometimes in life you just need pictures or data visualizations, like this...

    Or this: Number of wind turbines by state in the US?


    But this is Luxembourg...

    So we need finance examples...

    Example: Financial Analysis

    Example: Moving average of AAPL

    Risk is also data science challenge


    Especially with changes in regulation...

    But I don't work for a corporation

    Data can be used for NGOs too...

    Data can be used for NGO's

    This is a web app of house prices and commutes, done for an NGO in London who wanted to show the effects of
    changes in house prices on peoples commutes.

    Reducing Maternal Mortality Rates in Mexico

    Mexico - Presidencia de la Republica -

    The maternal deaths in Mexico from pregnancy, childbirth or postpartum complications have decreased from 89 deaths per 100,000 live births in 1990 to 43 in 2011. Despite this improvement, the rate of decline has significantly slowed and Mexico is not on track to achieve its Millennium Development Goal of reducing maternal mortality 75% by 2015.


    But what if you're in Politics?

    Earlier I showed how data can
    even predict 49 out of 50 states
    in the last American election.

    And that Obama would be the president!

    So why would I need a data scientist?


    You may already have one. I know numerous business intelligence, data analysts, business analysts, risk analysts who ARE data scientists.

    Alternatively you can hire a data analytics consultant to help you get started.

    But what signals should I look for?


    Well there are many answers... Like...

    Does this sound like you?

    Are you not taking full advantage of your reporting?

    Do you need a high level visual overview of your operations?

    Are you targeting your marketing efforts effectively by using the right customer segmentation - by age or gender for example?

    Or this?


    • Are you losing customers and not understanding why?

    • Are you making decisions on the basis of data or on the basis of 'gut feeling'?

    • Are you changing your websites or products on the basis of data driven experimentation?

     

    Example: What-if analysis...

    What about communication?

    Mathematically sound communication to clients: you may have situations where you need the data scientists to talk directly to clients or to their data scientists.


    This is yet another reason to make sure you hire someone with excellent communication skills, because they will be representing your business to really smart people.


    Data Scientists are like 'translators'

    A lot of my work at the moment is mathematical communication with external stakeholders and Professors.


    At Amazon a lot of my work was with Research Scientists in Optimization. Translating their ideas for business stakeholders.


    I often have to translate from the 'business' to the 'software' team.


    Do you have someone like that on YOUR team?


    If this sounds familiar then you might need to hire a data scientist!




    I hope the examples helped

    I hope it is also clear how data can be used in whatever field you work in.


    I hope it is also clear that 'big data' is not something to be scared of but should be part of your organizations strategy.


    I know that developing a data-driven culture is extremely difficult.

    Thank You For Listening


    Any questions?

    Reach out to me if you have any data questions.


    @springcoil
    Search Peadar Coyle on Linkedin
    peadarcoyle@googlemail.com

    DataScience

    By springcoil

    DataScience

    • 2,086