0:00

[MUSIC]

Â Let's go to some conceptual preliminaries

Â that would help us navigate the rest of the session.

Â So preliminaries on math modeling and

Â these are some basic questions that would arise.

Â Question one on math modeling, what is a model?

Â What is a model?

Â So basically, the word is used quite often and

Â we have some weak sense of what it means.

Â Let me define it for our purposes.

Â A model is a set of relations between variables of interest.

Â There are two variable types, broadly speaking,

Â there are dependent variables and there are independent variables.

Â If you want to think about an example for a model, think about let's say sales

Â being a function of the four Ps in marketing or the CAPM model or

Â industry profitability being a function of the five forces and so on.

Â So, you have this bunch of variables and we are trying to in some sense estimate or

Â put together a relationship between them.

Â Question two, any examples of traditional models from your daily work?

Â True, we use a lot of methods and they would count as models too.

Â Think about a regression model, a logit model, factor and cluster analyses.

Â Think about ANOVA, the analysis of variance and so on.

Â Question three, why care about modeling in analytics?

Â And why should we care so much about it?

Â If you recall the original definition of analytics, there were two paths.

Â So, getting from real world questions to real world answers.

Â One was experimentation, the direct path and the second was through math modeling.

Â And thereby, analytics.

Â We care about modeling in analytics,

Â because it provides us two major things, explanation and prediction both.

Â What are the characteristics of a good model?

Â Well, yeah, these are ideal characteristics and

Â there are very few models.

Â Seldom are these characteristics met in full.

Â So for instance, a model should be simple, it should be small,

Â it should be generalizable, it should quick to setup and run.

Â It should have high explanatory power, high predictive power, even with small

Â samples and then what are the odds of finding a model that does all of these?

Â But that in some sense is the ideal and it helps to keep the ideal in mind,

Â which brings me to modeling typology.

Â Remember, we talked about these two variable types,

Â the dependent variable Y and the independent variables X.

Â So for instance, there are two types of well, let's say,

Â marketing research model types.

Â Your modeling of dependence relations on the one hand and

Â modeling of interdependent relations on the other.

Â Modeling of dependent relations we had seen in the last session.

Â These are models of the kind Y is some function of f(X) where Y is

Â a dependent set of variables and Xs are the independent variables,

Â and three components of a dependent relation model are these three.

Â The dependent variable Y, the functional form f and the independent variables X.

Â We are not going here today where we are going today is that,

Â modeling of interdependence relations.

Â Interdependent modeling.

Â Why are we going here?

Â Sometimes, there may be no dependent model clear card.

Â I mean, there may be no dependent variable that comes out.

Â In such a situation, interest would center on exploring interrelationships between

Â whatever variables we have and all of them would be in some sense X variable,

Â independent variables.

Â Sometimes we also say, we are actually looking for

Â the underlying structure of this variable set.

Â For example, take a look at that.

Â That is a bill, basically, a shopping bill.

Â And what you see there, a set of products that were bought and

Â these items are part of the same basket.

Â And because they're part of the same basket,

Â perhaps there exists an affinity between them perhaps.

Â If I have millions of shopping bills, I can actually compute probabilities for

Â these affinities.

Â What are the chances that product A and

Â product B are going to be part of the same shopping bill?

Â Even better, if I know in some sense a customer unique ID,

Â things become even more interesting.

Â Now the question is in some sense, what are basically the similarities?

Â So in some sense the question now would be which customers are similar right and

Â their basket composition and so on.

Â So interdependence modeling helps us answer questions such as one,

Â what are the similarities and difference among our customers in our products and

Â service lines?

Â Two, how similar are these guys?

Â How similar or different are they?

Â Three, why do we need primary data?

Â I mean, why?

Â Why are they different or similar?

Â So, you might need primary data for this.

Â So that's one of the possibilities.

Â Which finally brings me in some sense here?

Â In what follows we are going to focus on three approaches to exploring,

Â understanding and

Â working with interdependent modeling on the customer side of business data?

Â What are these three things?

Â One, factorizing data.

Â Two, clusterizing and I don't know if there is such a word, but

Â you get the point.

Â Clusterizing data and three, visualizing it.

Â In addition, we will apply these three approaches to

Â an unstructured form of customer side data which is text and

Â subsets of it would be opinions and sentiment.

Â So, we will actually see all of these in action in the rest of today.

Â [MUSIC]

Â