Design thinking concept brings out the importance and integration of “Consumer Centric” product development, into the “Consumer Centric” marketing. Read, http://hbr.org/2008/06/design-thinking/ These principles are fundamental to product usability by product users, digging out the “preferences”, “usage”, and “ambiance” of the markets, so that the full life cycle of the innovation, consumers, and markets are integrated into the development process. – Designing thinking is “…to conceive of a fully developed marketplace, not simply a discrete device” – Design thinking is “… — a methodology that imbues the full spectrum of innovation activities with a human-centered design ethos.” – Design thinking facilitates a leadership commitment of “…a profession that blended art, craft, science, business savvy, and an astute understanding of customers and markets” – Design thinking “…is a discipline that uses the designer’s sensibility and methods to match people’s needs with what is technologically feasible and what a viable business strategy can convert into customer value and market opportunity.” I like the HBR article more than the video below, and it is fun to hear them how they are struggling to articulate…!
A quick intro here:
The following link is a great example how to use Issue Tree methodology on how to connect the research questions and hypotheses of an opportunity. These hypotheses are structural hypotheses, not statistical hypotheses in the first stage and they are converted into statistical hypotheses, subsequently based on the data. These structural hypotheses have to be evaluated and the one with expected maximum pathway wins.
It provides various hypotheses pathways and different pathways are evaluated on the expected pay off as follows. This is also from the same reference mentioned above.
A case study going through the tree details is given here on the issue of “how can we increase profit”, and hence the “issue tree” for “increasing profit”.
Now for each of the issue-tree end point, we can calculate
In an example based on more dimensions, look at
Please read all the wonderful details the first link provides.
The following is derived from partial information provided in the article “Machine Predicts Heart Attack 4 Hours Before”
This is a great example of how to calculate and interpret the various measures in the “Confusion table”. Find all the missing counts and their %s.
Interesting thing for me is that “The true positive” is 60% for a false positive of 20%. In this example, the authors used 72 variables to create the algorithm.
Isn’t it confusing that the sum of True positive rate and false positive rate is not 100% (here it is 80%)? Why?
Do you know how many samples were exhausted before you get that 60% true positive?
Titles are enhanced for better understanding of contexts. Competition projects are done by multiple groups. In such cases, a random selection of a group is chosen to show the structure of project developments. Some times, only a key point is highlighted in the title.
Health Insurance Analytics – Recommendations to Increase Enrollment For the Affordable Care Act
Group: Derek Hulley, Dave Jacobs, Drew Moore, Greg Wolford
Insurance Analytics – Predicting Most Favored Coverage Options for an Auto Insurance Company – Allstate Kaggle Competition
Group: Amit Bhagwan, Ian-Sean Caven, Patrick Sawyer, Joshua Weiner
Sports Analytics – Redefining the NFL Player Value Chart
Group: Krystel Kearns, Sean Liddy, Hui Liew, and Tom Robinson
Creating Recommendation Engines With Yelp Data – Recommendation Engine for Recommending Restaurant for Users and Recommendation Using User Response Data and Item Response Data
Group: Brian Hughan, Andrew Angliochetti, Michael Redmond
Hospital Analytics – Predicting Sepsis
Group: Jennifer Su, Erin Sahlsteen, Lauren Shores, Mark Schumacher
Web Analytics – Moving Sales from Brick to Click
Group: William Hancock, Stephanie Rosenberg, Mora Ambrey, Indu Sriram
Online Retail Analytics – Predicting Merchandize Returns (Prudsys Competition)
Group: James Braun, Peter Golden, Kathryn Turnbull, Brian Young
I take pride that I can lift up my students to ask the right questions, and sincerely follow through the various answers they may get and evaluate them systematically. Some times I say, good well formulated questions, are the answers, until you get to the next level.
Sometimes people get threatened by the mention of “Ph”. Democratization of opinions have been well instituted by social media, rightfully, and any one can make their contributing comments, notes, and all kinds of interpretations.
Here is a fair balance statement, I wrote some time back about having a PhD vs. not. You may enjoy reading to the end.
“You can define what it is or what it’s not, you may defend it or offend it you can say without its peripherals it is nothing, or it is the only thing and peripherals are nothing without it, you can put a question and make others follow to your den, or you can lead and provide your answer. You may have phd and foul it or you can be any thing and still foul it”
– linkedin business analytics grp 7.5.09 (the original posting had a typo, “fout” instead of “foul” at the second time, and interestingly it is also valid)
Be confident my friends. Only good questions, and sincere accommodation of scientific answers matter and nothing else matter. In between, we see showmanship, marketing, and money making, and that is also fun. Life is a carnival often times.
The light does not require the rooting from darkness. – Missed notes from Yogi Bera.
Please indicate your choice subjects when you send me your resume at nsambamoorthi at gmail dot com.
1. Regression and Classification
2. Time Series and Forecasting
3. Machine Learning
4. Multivariate Methods
5. Sampling and Design of Experiments
6. Mathematics for analytics
7. Business Statistics
8. Marketing Analytics, …
reviews in a simple easy to understand writing, the various prominent algorithms.
The abstract reads, “Supervised machine learning is the search for algorithms that reason from externally supplied instances to produce general hypotheses, which then make predictions about future instances. In other words, the goal of supervised learning is to build a concise model of the distribution of class labels in terms of predictor features. The resulting classifier is then used to assign class labels to the testing instances where the values of the predictor features are known, but the value of the class label is unknown. This paper describes various supervised machine learning classification techniques. Of course, a single article cannot be a complete review of all supervised machine learning classification algorithms (also known induction classification algorithms), yet we hope that the references cited will cover the major theoretical issues, guiding the researcher in interesting research directions and suggesting possible bias combinations that have yet to be explored.”
A great reference. Go for it.