Big Data Business Analytics

Analytical Projects

For several businesses, I've worked on analytical projects. Below are some demos and previews of my work.

• Quantitative Customer Journey Predictive Analytics

• Display Banner Rotation Optimizer

• Behavioural Customer Trend Anomaly Detection

• GasShipping Optimization Model

• Value At Risk Energy Trading Model

• Commercial Spot Frequency Value Calculator

• Movie Scheduler

• Data Point Editor

• PurpleSpy++ Computer Security Monitor

Quantitative Customer Journey Predictive Analytics

Programming language

Teradata / SAS / MicroStrategy / Celonis Process Mining

Operating system

Unix

Release dates

2015 (MVP), 2016 (MicroStrategy Report), 2017 (Predictive Module), 2018 (Process Mining Module)

Type

1-on-1 Marketing forecasting model

Description

The Quantitative Customer Journey Predictive Analytics big data tool developed for a major financial institution in The Netherlands, enables the 1-on-1 forecasting of retail marketing campaign effectiveness and smart target audience selection. The big data tool running on Teradata platform daily tracks customers and prospects end to end in their cross channel journeys for buying, using and retaining financial products like savings accounts, loans, residential mortgages and investments. This custom built Customer Journey suite runs nightly in order to provide next-day business intelligence insights for retail marketing campaign effectiveness using MicroStrategy dashboards. Behind the scenes the forecasting module implemented in SAS Enterprise Miner automatically optimizes running campaigns for net journey effect, such that campaigns will be automatically adjusted to the most promissing audience with regard to sales and service. Finally, the opportunity insight component discovers and signals previously hidden marketing opportunities across channels and products.

Display Banner Rotation Optimizer

Programming language

Teradata, MicroStrategy

Operating system

Unix

Release dates

2017

Type

Display Banner 1-on-1 Reach Planning Model

Description

The Display Banner Rotation Optimizer consists of a first customer response eliciting mechanism based on group similarity characteristics of the client. By optimally rotating random display banners with a higher than average probability of click-through for the comparison group, online customer interactivity for retail clients of a major Dutch financial institution was increased. The statistical model behind the banner planning system was quite simple in mathematical terms, but very valuable for business value generation as no longer the default banners were served to customers. The improved approach of banner rotation resulted in a higher reach and better response on the on-site bannering channel. The results were monitored in a MicroStrategy dashboard of the marketing department.

Behavioural Customer Trend Anomaly Detection

Programming language

Teradata / SAS

Operating system

Unix

Release dates

2017

Type

Trend Anomaly Detection model

Description

The Behavioural Customer Trend Anomaly Detection developed for a major financial institution in The Netherlands, automatically tracks any significant changes in many customer behavioural attributes like cross sell, channel usage and financial status. The big data tool runs on a monthly basis on the SAS Unix platform, producing actionable signals marking customers whose behaviour trend has changed above/below a certain personalized threshold. Allowing the financial institution to monitor millions of customers on changing patterns in behaviour and attributes, such as for purposes of acquisition, retainment or churn. The software monthly-refreshed results are easily imported into other tooling such as marketing automation and fraud detection monitoring systems.

GasShipping Optimization Model

Programming language

MATLAB

Operating system

Windows Server 2003

Release date

2008

Type

Financial computation model

Description

The hourly gas shipping allocations for the energy firm needed to comply with the regulations of the Network Supervisor. Also the allocated capacity may not differ too much from the real used capacity, otherwise penalties would be given to the gas shipper. A VBA-model, with lots of sheets and formulas, was turned into an easy to maintain automation script. Computation time was improved significantly and several gas portfolio's could be calculated and optimized at the same time. Computational results were both exported to a spreadsheet as well as understandably presented in graphs and charts for management. With the newly developed software, more savings could be achieved on avoiding penalties and by better pricing.

Value At Risk Energy Trading Model

Programming language

MATLAB

Operating system

Windows Server 2003

Release date

2006

Type

Financial computation model

Description

Using the mathematical programming language a financial model was turned into a automation script. With the Energy Trading Model the daily financial risk positions for the commodity portfolio (Energy, Gas, Oil) could be calculated. The thousands of Value-at-Risk positions were calculated in mere seconds, and when the model was done a user interpretable graph was displayed to give an overview of total risk. I.e. green is good low risk, red is bad high risk.

Commercial Spot Frequency Value Calculator

Programming language

Java

Operating system

Windows XP

Release date

2006

Type

Econometric optimization model

Description

Software written for my Master's Thesis that implemented a mathematical model for computing the Reach and Frequency for television spot time slot programming. The program loaded a commercial break schedule as well as viewer probability matrices (e.g. retrieved from a rating agency) and then outputted the best scheduling chances in XML.

Movie Scheduler

Programming language

C++ / VBA / Java

Operating system

Windows XP

Release date

2005

Type

Econometric optimization model

Description

Software written for my Bachelor's Thesis that implemented a mathematical model for allocating cinema movies to time slots in order to maximize ticket sales. The algorithm included the use of ILOG CPLEX software for column generation and branch and bound Lagrangian search. The output was written to XML and loaded back into a spreadsheet for management using VBA-code (1^st screenshot). In a second iteration I improved the Graphical User Interface (GUI) to a Java interface (2^nd screenshot).

Data Point Editor

Programming language

Java (Niagara Baja Framework)

Operating system

Java embedded systems

Release date

2004

Type

Automation and productivity software

Description

The Data Point Editor is a GUI-tool for automating climate control embedded systems programming/configuration tasks. With the Data Point Editor building automation project manager can easily connect climate control devices like heaters, vents, lights and motion sensors, with software logic and room switches. Team productivity increased, projects were finished earlier and more business customers could be served. Over the years other colleague improved the Data Point Editor further.

PurpleSpy++ Computer Security Monitor

Programming language

Visual Basic

Operating system

Windows 95

Release date

2000

Type

Security software

Description

The application was designed to protect the computer lab environment for misbehaving high-school students. PurpleSpy++ runs on the background and monitors unusual activity by the user. Observations are reported in a log for the IT-staff. In a sense, the program made it possible to have Guest Users at your Windows 95 computer, without them damaging or removing your files.

Quick Start Guide to Advanced Analytics

In this section you'll quickly get up to speed on the topic of Advanced Analytics. This will help you known when to hire or look for an Advanced Analyst (Econometrician, Data Scientist, Business Analyst, etc.).

• What is Big Data Business Analytics?

• Why Big Data Business Analytics?

• Big Data Business Analytics specializations

• Big Data Business Analytics university degrees

• Companies using Big Data Business Analytics

What is Big Data Business Analytics?

Big Data Business Analytics means analysing big volumes of complex data for systematically optimizing business goals.

• It's so much data, that you cannot use a regular calculator or spreadsheet program any more!

• The data can be both structured and well formatted, as a complete chaotic pile of unordened mess.

• And there are so many (inter)dependencies between the data, that you cannot find the connections and relations by hand any more.

How on earth are you going to analyse this big data for getting actionable insights?!?

Luckily, there are solutions for exactly that!

That is what a Big Data Business Analyst, Data Scientist or Econometrician helps you with.

Steps in analyzing and using big data

1. Formulating an accurate problem description (often a hypothesis for a business case)

2. Retrieving many data sets that may contain information on the cure for the problem

3. Ordering and preparing the big data sets for easy scientific analysis

4. Modelling the hypothesis as a statistical problem description to solve

5. Crunching the numbers in data science software that tries to find relations between your hypothesis and the ingested data

6. Evaluating the accuracy of the model outcome of training set data on test and validation data sets

7. Presenting your findings to your stakeholders

8. Coming up with a general statistical approach that can be productionalized at scale

Why Big Data Business Analytics?

Big Data Business Analytics (or Data Science, for that matter) isn't new. Actually, this profession already exists for decades. It is also known as econometrics and management science. So Big Data Business Analytics isn't something new. It's just a long-time proven fact-based approach to structurally achieve higher business goals, with a new trending fancy name.

However, due to the massively increased computing power over the last 10 years, complex business problems can now be solved within an acceptable amount of time using scientific techniques and advanced mathematics. The smart phone you now have in your pocket or purse contains more technology and has more computing power than a space station. The same holds for enterprises, that now have access to sophisticated big databases and worldwide computing power. In other words, business now has the tools to compute the answer right when it is demanded.

At the same time, due to the digitalization of the world, the pile of data increases more and more rapidly. In essence, there's so much data worldwide, that you'll never have enough time to organize, understand and utilize all that data. And we as a society keep adding data to this stock every day, so the amount of data doesn't stop to grow. Therefore smart computer algorithms are needed to help business find the right answers within that (sometimes messy) data.

So, because nowadays more and more people are acquainted with more data and faster devices every day, we get to understand that Big Data Business Analytics is more than before needed to achieve business success.

Big Data Business Analytics specializations

The business goals can be in the area of corporate finance and risk management, marketing intelligence and strategy, operations and supply chain management or general macro/micro economics.

• Corporate finance and risk management aims to manage company finance in a global enterprise scope. At the same time huge numbers of outstanding loans, options, bonds and other financial instruments need to be carefully analysed, valued, assessed and monitored. The main goal is to find ways to make a better return on financial investments while carefully considering risks like counter party risk, risk of default, market risk and risk of (de)valuation.

• The goal of marketing intelligence and strategy is to really understand the customer and discover how products and services can be best presented to the customer 'en masse', while keeping the customer experience personal and tailored to his/her needs. The main goal is to create and keep customers for the company brand, and sell the products & services for a profitable price.

• Operations and supply chain management is applied in areas such as logistic networks, transport & shipping, inventory control, route and schedule optimization. It can also be applied to make systematic processes within a company more efficient and productive. The main goal is to save money on operations and make company processes work fluidly together.

• You'll find general macro/micro economics mainly in governmental and political institutions and research laboratories. Mathematical and statistical techniques are used to find relations and segments within nation wide economic topics, like demographics, geography, politics, warfare etc.

Big Data Business Analytics university degrees

In The Netherlands (the country I live in) there are 6 universities that teach econometrics. Some of them have labelled their courses Big Data Business Analytics. Other educational institutions call it Quantitative Research or Data Science.

Top notch education is where the areas of business administration, computer science and data analytics are combined. Preferably with the entitlement of a Masters Degree in Science and Business Administration once successfully completed. Why? Because of this simple formula:

Big Business Targets × Scalable Enterprise Technology × Fact-Based Continuous Improvement = Lasting Enterprise Success

Programmes can be followed both full-time as well as part-time. The part-time option is very interesting for people who already have a Master's Degree and a solid quantitative background and relevant work experience. Be sure to look out for education where theory and practise is combined: learn & apply.

Typical courses that are taught are:

• Business Courses

  • Competitive Strategy
  • Consumer Behaviour
  • Corporate Finance
  • Entrepreneurship
  • Financial Accounting
  • Financial Modelling
  • HRM
  • Law & Ethics
  • Marketing Strategy
  • Operations and Supply Chain Management
  • Risk Management
  • Valuation

• Analytical Courses

  • Statistics
  • Econometrics
  • Decision Models
  • Quantitative Marketing

• Computer Science Courses

  • Artificial Intelligence
  • Computer Infrastructure
  • Databases
  • Data Mining
  • Data Visualisation
  • Machine Learning

• Leadership Courses

  • Business Cases
  • Field & Study Trips
  • Leadership / Management Development Programmes

Help with study choice for scholars

If you're a scholar interested in obtaining an university degree in econometrics in The Netherlands, be sure to visit www.econometrie.nl as this website contains lots of information in Dutch about studying econometrics in The Netherlands. The website is maintained by Stichting Landelijk Orgaan der Econometrische Studieverenigingen (LOES), where the six student associations in econometrics are united.

Help with traineeships and starter jobs for graduates

The student association that I was member of during my study Econometrics at the Erasmus University Rotterdam, maintains the website Econometrie.com, a nice website with listing of traineeships and starter jobs for young professional econometricians.

Books on Big Data Business Analytics

Recently, I've reviewed the book Creating Value with Big Data Analytics by Peter Verhoef, Edwin Kooge, Natasha Walk for ManagementBoek.nl and my professional contacts at Platform voor Klantgericht Ondernemen. It's an excellent English book on Big Data Analytics, targeted for both managers and data specialists, written by Dutch experts.

• Click here to read my review (Dutch)

Companies using Big Data Business Analytics

Because Big Data Business Analytics is so hot, there are many companies that are using Big Data Business Analytics. At the same time these companies regularly have vacancies for both junior and senior positions in Data Science and/or Econometrics.

Hiring an Econometrician is not cheap, therefore mostly big companies with big budgets (the so-called Enterprises) hire these people. Have a look at Quintus' Resume to see what companies I've worked for thus far.