Monthly Archives: November 2014

Happy Holidays and Season’s Greetings…Get the best opportunity analytics done for your organization

One of the most wonderful times of a year.   Be with family, play games with friends, family, and kids.  Make promises for the new year and laugh about it even while you are making them.  My best wishes to my students, clients, and every one who is visiting my site.  Happy Holidays and Season’s Greetings.

Season’s Greetings

Season’s Greetings.HappyThanksGiving

Do you want to access and utilize the power of geo-demographic, preferences, and life stage data of 100 plus million adult consumers along with your internal data for marketing purposes using multi-class and multivariate advanced analytical methods, for

–          Acquisition of new customers to your products and services

–          Cross-selling

–          Up-selling

–          Partner complementary product selling?

Integrated CRM analytics utilizing the rich internal data, third party geo-demographic/preferences data, and survey data, is what we do.

Our team is supported by world class experts who are both experienced in doing hard core modeling best practices as well as consultative solutions to your needs

– We use statistical and machine learning methods and AI procedures

  • For product, brand preference models, and media channel preference models
  • Healthcare outcome research models
  • Right person, right message, right offer models
  • Integration of web data with any of your marketing data for marketing optimization

–          We also provide dashboard for your marketing execution

Please reach out to me to provide you more details on how you may benefit with our works

With warm and best regards, and season’s greetings

Happy ThaHappyThanksGiving2nks Giving.


“Sam” Nethra Sambamoorthi

======================What do we deliver ?=====================

Predictive Modeling is all we do.  We are data intelligence and analytics engineers.


Marketing Analytics

–  Product preference, brand and media preference predictive models built as solutions

–  We serve financial services, investor services, insurance, sales agents/advisors solutions

–  Integration of geo-demographic, preference, social network, and lifestyle and lifestage data with sales activities of sales people, web advertising

–  Integration of geo-demographic, preference, social network, and lifestyle and lifestage data with sales activities of financial advisors

–  Integration of geo-demographic, preference, social network, and lifestyle and lifestage data with sales activities of insurance agents

– Integration of geo-demographic, preference, social network, and lifestyle and lifestage data with web data

–  HR analytics

  • Likely to move, likely to start their wireframe shops, likely to join competitive company, likely to become independent
  • Value of an employee
  • Personality types of employees
  • Risk individuals

Some of the best complements I have received, as analyst expert:
“… You delivered a solution that is most effective in the last 30 years …we have never seen anything like this before” – Head of Analytics, Top Investor Services Company. The client accomplished the yearlong project of acquisition in two months, along with two top accounts worth $500 Million AUM.

“… You solved a critical problem which our group did not …” – Head of Analytics, Top Technology Company

“… wish I can clone your talent …”, Head of Analytics, Top consulting company

“…I never thought that the data would provide such powerful insights… great contributions to our initiative…” – Project Manager, Top Credit card company

I am also thankful for the colleagues who complemented my work on the account management side.

We also build complete web store platform solution with the following: 

– web setup

– web marketing

– web campaigns

– web payment gateway

– web analytics, and

– analytics intelligence reports for continuous improvements 

Pictures are from clipart of

Some Educational Excel Charting Videos

Multiple bar graphs in Excel

Multiple line graphs


Creating a percent and labels and stacked bar chart

Creating combination charts

Watch out the second part where secondary axis is used.  This is a confusing chart functionality and recommend not to be used.  You can always use multiple series data with the same horizontal axis.

Creating speedometer chart in excel

Drawing a pareto chart

Fishbone diagram for cause and effect (another one)

Gantt Chart (useful for project planning)

Simple control chart (a process is under control or not; usually used in statistical process control)

What is Wrong with This Microsoft Graph with Additional Axes?

Microsoft in its website on “Add or remove a secondary axis in a chart“, provides the following as an example of how to add additional vertical axes to the graph.  Here is the copy of the graph.

Formatted combination chart

People are likely to interpret the relationship between the two important elements of the graph and likely to loose the relationship to month in the process.

When the average price of home is $400,000, the number of homes sold is 100,000 or when the average price of home is $430,000, the number of homes sold is 25o,000.  Since the logic can go the other way also, as there is no price elasticity is implied in general discussions, one may say as, the number 100,000 homes sold for $400,000  price and 250,000 homes were sold with the average price of $430,000, as points of interpretation.

However, the actual data from the same page is

Homes Sold Average Price
Jan 280 410
Feb 150 450
Mar 220 430
Apr 275 425
May 155 410
Jun 255 400

This is a time series data, and the simple time graph provides the following.

The graph clearly points, no correlations to implicate any interpretation.   This is the right way of representing data for visualization.

Do not use multiple axis in a graph.  Use appropriate graph.

Classic Developments in Data Mining, Predictive Modeling, and Visualization

Today, I had the joy of seeing the following, Knowledge Discovery in Databases, An Overview by William J. Frawley, Gregory Piatetsky – Shapiro, and Christopher J. Matheus, an awesome find and traces back how giants in the field think and what happenstance saw opportunities that run of the mill statisticians do not get the chance to see and solve problems.  Definitely, this is number 1, and it starts the revolution in the thought leadership and tools for expansion of data analytic methods.  I may not be able to say the ranking for other great references.

This perked my interest to locate some such references, that every analyst – data mining, data sciences, and knowledge discovery students, information strategist, should be keeping in their library.

So I decided to locate top references that every student of data mining and predictive modeling scientist should read about or at least get the best balanced knowledge (“The sum or range of what has been perceived, discovered, or learned”, – I like this as most subtle of all the definitions I have seen).  The details of subtleties are expanded in the definition provided in

I encourage you to buy the book.  There is a wealth of information on how different people brought together these ideas early in the game.

The second reference is

by LUKASZ A. KURGAN and PETR MUSILEK (2006, Knowledge Engineering Review, Vol 21:1, 1-24, Cambridge Univ. Press) – “This survey presents a historical overview, description and future directions concerning a standard for a Knowledge Discovery and Data Mining process model” and additionally adding new developments on the importance of interoperability and automation of the process, a precursor for more and more acceptance and development of machine learning approaches.

In terms of methodology, the reference, Top 10 data mining algorithms, published in Knowl Inf Syst (2008) 14:1–37, that became a book later on, because of great interest in these algorithms, is a must reference for data mining/predictive modeling community.  The authors of this is a long list of distinguished people in the industry: Xindong Wu, Vipin Kumar, J. Ross Quinlan, Joydeep Ghosh, Qiang Yang,  Hiroshi Motoda, Geoffrey J. McLachlan, Angus Ng, Bing Liu, Philip S. Yu, Zhi-Hua Zhou, Michael Steinbach, David J. Hand, Dan Steinberg.

The 2012 survey of 798 people who answered questions related to what software they used in the last 12 months, it is amazing to see EXCEL ranking second.   The tabulation and graph that includes the red color high lighting for open source applications is a great summary, however, biased potentially this may be.  The article does not discuss any possible bias here.   See for more interesting and useful details. publishes regularly updates with surveys such as this. should be one of the go to resources for any data mining enthusiast.

Kdnuggets also did a survey that published recently, few months back, on the topic of “What programming/statistics languages you used for an analytics / data mining / data science work in 2014?

R, SQL, SAS, Python ranked on top.

In case you are interested in knowing more details specific to top six open source data mining software/applications, look into

Different Types of Analytics Reports – A Review1

In the early stages of analytics revolution people structured different types of analytics on the basis of the following models; information reports, analytical reports, and proposals.

Informational reports offer data, facts, feedback, and other types of information, without analysis or recommendations.

Analytical reports offer both information and analysis, and they can also include recommendations.

Proposals offer structured persuasion for internal or external audiences.”

The quote above is from the educational training article on the above understanding here.

Also, read the comprehensive notes on business writing here.  This may be skipped by 402 Class. .

Realizing the importance of creating ‘All Possible Summary Reports’ and with improved technology, reporting systems such as “Crystal Reports” started getting more sophisticated than the above categorizations, supported by OLAP and Summary Information Cubes.  Subsequently it is owned by SAP.  Now such functionality is available in every software giant. Oracle, Microsoft, and

Subsequently, the model of analytics evolution, analytical thinking, and creation of reports was posited on the basis of the paradigm “Data, Information, Knowledge, and Wisdom”.

This gives more weight on what the contents are with lofty goals with words like “knowledge” and “wisdom”, and creating reports that are representative of knowledge and DifferentTypesOfReportswisdom were either reduced quickly to interpretation of summary reports or too high a cynical goals to achieve.  A comprehensive summary is available here.

Recently, the paradigm focuses on tactical reports detailing on the evolutionary approach to creating and building business intelligence on the basis of the paradigm, “Descriptive, diagnostic, discovery, predictive, and prescriptive”.   A fun explanation of this is provided here.

So depending on the analytics maturity level of your organization, you will be writing analytics output reports that are reflective of the analytical activities that are associated with with your organization.

Typically reports created out of sample surveys are descriptive  reports.

Diagnostic reports look for spikes (outliers), correlations, and associations, but it is just the start of predictive and prescriptive solution.   Here is an example of how keyword performance on the search network is diagnosed for treatment, that Goolge uses for delivering web analytics report  These are any thing short of detailed work on modeling that imply causation and prediction.

On the other hand, Predictive reports heavily use scoring methods that help rank high value vs. low value units or consumers, using different weights for different variables, using the predictive models.

The key page in the predictive analytics report is the lift table and all the supporting document that go with that; for example, data discussion, variable importance, and methodology discussion are the supporting documents.

Here is an example of a lift chart.

The idea of how and where prescriptive part of analytics interface with predictive part is explained here.

If predictive modeling is done properly with proper sampling and design processes, there is no need of what if situations, one would imagine.  While every previous step is an improvement over the earlier one, the prescriptive part is not an improvement but a way of applying the predictive model.  These what-if scenario type simulation studies will help to see where the budget needs to be allocated and how much, for example.

An example of prescriptive reporting is provided here.  Accordingly, if simulation based, “what-if” analysis related outputs are not discussed, we truly do not have any thing deep differences between predictive and prescriptive in general.  However, there are simpler situations where predictive models and the weights of the important variables in such models can provide prescriptive solutions.

OLAP cube from:

Report image from:

Measurements and Data Requirements – Analytics Layered Business Requirements Document

The best place analysts need to be and should ask the management permission to be part of the mission of bringing analytics to an organization is at the time of formulation of new projects or initiatives in an organization.

All these are captured in the business requirements of a project and some times, also called strategy document.

As a start read Business Requirements Document: A High-level Review to understand the general principles of what it is and how it captures various aspects of a project.

Your duty:   Ask clearly for success measures.  Remember everything starts with what, why, how of planned  measurements in any project or initiative.

Hold on to the success measures; that is your magic carpet that will take you around where you want to go.  You, and you are the only one who can provide the analytical strategies to achieve the movement in the favorable direction of  strategic metrics, divisional or overall, and validate whether weather various success measures support the department’s strategy and hence the corporations overall strategic metric.  The management usually understands the your duty is to make sure the organization is “datawise” and all processes are measurable, and can be supported and proven with data.

Ask analytics section to be introduced in the strategy document, if it does not have one and if it has one, make sure it will achieve the project purpose, and as always provable by appropriate data capture and usage.

Remember, do not get too geeky in the section regarding all the smart works of a statistician or a computer scientist.  Managers trust you will take care of it, but are very happy to understand that you are thinking on how to make sure their work will be received and judged properly.  Work with project owner.

In terms of business requirements document, there is a way to set the clarity at many levels.

Here is a reference that helps understand the business grammar to achieve the requirements.

Business grammar should be supported by clarity required in the definition of variables captured and their interpretation and usage subsequently in the analytical steps.

Here is a generic document from academic training point of view.

Business Requirements and Data Requirements.

Here is well laid out template document prepared by the British Columbia government which provides the training on how to write a business requirement documents but also beautifully connects to “Data requirements” in a business requirement document.  See here.

Remember, all the interactions for data requirements and figuring out the measurements need to be driven by the “consumer centricity” approach, meaning the input, experience, and design thinking that go in to the business requirements are meant to leverage the consumers as the center in all the business processes.

Consumer Centric Balanced Score Card Method of Performance Management

The balanced score card is a well known concept.  The columns address the goals, metrics, and processes that are essential to keep the focus on multiple projects/initiatives and thereby balancing, prioritizing, and keeping all the happenings in an organization connected, via all the tools and processes in (1) financials, (2) innovations, (3) internal processes in prioritizing, (4) learning and strategizing the differentiation

The moment you define this by segment by segment of your end users we will see a better focus on your balanced score card from the consumer centric point of view.

It is not a difficult concept but the spin is a very different angle/game compared to what is traditionally known as balanced score card.

In the following discussion I will also bring out why or how this may have a different angle in huge companies, where the under current is that, “well our organization is really serving the whole population”.

My estimate is that at least 20% waste could be eliminated if the company has consumer centric approach, even in the largest organization in the world, and also even big non-profit organizations as, the government.  This is a separate topic.

I am coming back to my main topic.  How to view and develop balanced score card when you commit your organization to consumer centric approach.

While performance management is fundamental, it could be based on any number of metrics.  The performance management has two functions to achieve, (1) to make sure multiple goals of various business divisions are supported until they keep converging on the overall  vision of the company, and (2) to make sure the performance metrics are getting achieved, in each of the key operations/groups/divisions.

The great example is PepsiCo. It is in more than 150 countries, and has several hundred brands. While this case in discussion could be an extreme example, because this is truly one of the largest organizations in USA and it dominates the beverage and snacks business, it brings out some important concepts for discussion.

PepsiCo is a 100+ years old company and it is in the midst of a consumer’s daily consumption of beverages and snacks. Their research center for innovations on tastes and food packaging is fair to say ahead of the taster.  So PepsiCo is an interesting company that is ahead of the consumer, to predict what the consumer will like or not like.  It is similar to Apple, but in the consumer beverages and snacks business.

So technically, I can speak about PepsiCo and its performance management and why it is following the balanced score card approach.  These insights are based on the various interviews by PepsiCo executives, news articles, and the principles of consumer centricity.

These perspectives are not about tolerable ups and downs in ROI and ROE as they are also due to many other macro economic situations and life style trends, and hence, it can not be in the interest of the company driven by ROI only.

In the same way, it can not be about cost-benefit analysis.

Further more, it can not based on IRR or discounted cash flows, because it’s product is consumed every day.

It has many social programs such as “Pepsi Fresh”,  on water conservancy, reduction of plastic usage, reduction of energy consumption, … , and so on.  These things can not be simply calculated as a direct ROI contributors.

So PepsiCo management uses what is called Performance with Purpose balanced score card approach.

Oakland A’s does not require a balanced performance management scorecard – think about it.

“A Consumer Centric Strategy Mapping Method that Connects Strategy, Innovation, Consumers,…”

The linkage map is all about how the different factors in each of the four consumer centric activities across the organizations are linked.  For illustration purposes, I sketched as 3 factors per line item but it could be very easily a varying number in each of the four line (row) items under consumer centric operations.  Each of the ovals represents your organization’s specific strategy idea/concept. These are grouped by key operating departments/groups.  There is not hard and fast rule about the number of such operating groups, though I have high lighted as three for each of the groups.

When you create this for each of the segments of your consumer population, you will see the pattern as to whether your consumer centric approach is coming out as planned or not via a balanced score card approach.  A functional way of interpreting “balanced score card” is that it is  a balanced view of the needs of “are all cylinders are working in unison, to fulfill the broad vision/mission of the organization”.

The structure of the views of the balanced score card will be a cube (as in data summary cube) where the third dimension represents various customer segments.  This is the reason why every organization should have a strategic segmentation scheme of their customers/consumers.  Ask your senior management for your organization’s strategic segmentation scheme.  The best way to bring that out in conversations is to connect that to the dashboard you are working on for the management.

Also read, why BSC fails,

The above critical article is too negative but it is good to have a bit of salt.

If a company is in the early stages of growth, and started growing in many angles, the balanced score card methods with the ideas of “consumer centricity” is a powerful approach for prioritization and synchronization of all the business processes.