2: Consumer and Consumer Analytics: Analyzing and Predicting preferences and Choice Flashcards
What does Utility consist of?
Utility composed of observed and unobserved attributes.
Observed attributes: Flavor, availability, Price
Unobserved attributes: Mood of consumer and decision maker, Personal context
Two forms of Brand and Product choices?
Brand and Product choice can be
- multinominal: Multinominal Logit Models, IIA property
- binary: Yes or no –>Logistic Regression
What is the Goal of Brand choice models?
goal is to predict the purchasing behavior and to model the process by which consumers make decisions
Why are linear models not appropriate to apply for Brand and Product choice Models?
Because Linear models assume that the disturbance term is normally distributed
disturbance term: is the (unobserved) value that are considered randomly
–>are not appropriate because they give linear continous values
However the observed output are binary (0 or 1, buy or not buy)
What does choice modelling estimate?
Choice modelling estimates the probabilitites (buying and not buying) and determines how they are affected by the observed attributes of consumers choice
What is the Multinominal Logit Model?
is a choice model where the consumer chooces between j alternatives
–>Here as well as in the binary case we assume that the alternative which yields highest utility is chosen
What is the Independence of Irrelevant Alternative (IIA) Property?
Multinominal Logit Models
Multinominal Logit Models suffer from the IIA, this property states
that the odds of choosing one alternative over another are constant regardless of whcih other alternatives are present
Why is the IIA property not realistic in marketing?
- In many marketing applications this is not realistic especially if some alternatives are close substitutes
- If similarities across alternatives are incorrectly assumed , the estimated effects of marketing variables are incorrect
How to deal with the Independence of Irrelevant Alternaitves
- perform statistical test of IIA e.g. attraction model
- use choice model that explicitly accounts for the fact that consumer choice behavior is affected by the composition of the choice set
–>Other models that eliminate the IIA assummptions are the Nested Multinominal Logit Model and the Multinominal Probit model
What do Markov Chain model?
model behavior over time and
consists of several observable (behavior, corresponding profits/losses) and unobservable (probabilities) factors
model how sequence of observations is related to transitioning among states
Markov Chain models define set of customer states based on…
The idea is to define the set of customer states based on:
- observed customer properties, e.g. the purchase of a product
- estimation of transition probabilities between the different states
- corresponding profits and losses
4 Probabilities of Markov Chain Models
- Transition probability: probability of going from one hidden state to another
- Emission probability: probability that observations are emitted from the hidden states ( prob observed outcome given particular state int he system
- Prior probability: basic likelihood for hidden state (prob that a random customer has a preferecne for brand A or B)
- Posterior probability: likelihood for hidden state given observation (prob that a customer prefers Brand A or brand B given that he buys brand A)
What is the goal of Markov Chain Models?
Capture….
Our goal is to capture dynamics in customer behavior over time, for
example, to see how firm interventions influence this behavior
How to infer prefercenes
(Markov Chain models)
Multiply every probabillity along the path (Viterbi Algorithm)
–>Machine learning system which identifies path with highest probability to be chosen
- finding most profitable preference combination
- If preference probabillities are close to each other advertising makes sense
For what do we use Logistic regression?
is used to predict binary or categorical outcomes (buy or not buy)
–>are binary choice models*
What are the two variables in logistic regression?
depdent variable: which is the variable of interest that we want to predict or explain
predictor variable/independent variable: are the variables used to explain or predict the depdent variable, these can be categorical but can also be continous (numerous)
What is the transformation applied in Logistic regression?
The logistic regression model applies a transformation (the logistic or sigmoid function) to the linear coombbination, whiich maps it to a probability between 0 and 1
–>This allows us to estimate the probability of the dependent variable belonging to a particular category or class
What are Logit Choice Probabilities?
Which method is used to estimate the parameters?
refer to the estimated probabilities of selecting a specific alternative among a set of available options
–>derived from the estimated coefficients of the chice model (logistic or Multinominal)
method used: Maximum liklihood methods
What are Purchase quantity Models?
models used to analyze and predict* the quantity of a product or service that customers will purchase.
aim to understand the relationship between various factors and the volume or quantity of purchases made by consumers.
What are the two utilities behind the Purchase Quantity Models?
For a set of J brands, ** 2*J utility equations** exist, which are arranged in two blocks:
- Utllities that determine the probability of brand purchase (Brand)
- Utilities that determine the purchase quantity (Quantity)
In which context are discrete continous models useful?
Discrete-continous models
- allow to solve multiple discrete continous problems jointly through one global utility process
- can e.g. be used to predict the purchase of a basket of yohurt flavors (both the probability of purchase and the expected purchase quantity)
How is the variable in Purchase quantity models called and what is its distribution?
the focal dependent variable is a so-called count variable that can take on nonnegative interger values (so not 5,3939)
–>This count data is distributed according to a poission distribution
What is the intuition behind the Poission process in Purchase quantity models?
the distribution of the number of units purchased in any interval depends only on the length of the interval
The random variable denoting the number of units purchased by consumer follows a Poission distribution with parameter llamda
llamada= how often does purchase occur
What do Duration models deal with?
Duration models deal with the duration or timing variables:
- time between sending and mailling and response
- Intepurchase time
- time between introduction of product and the adoption of user
What do Hazard Models do?
(Duration Models - Purchase timing)
account for censored duration variables. e.g. customer are unaware of events…
left censoring: events prior to observation period
right censoring: event that takes place after the observation period ( e.g. observation right–censored customer will never purchase the product or churn but only in the given observation, thus it might happen in other periods)
What is the advantage of right censoring?
Right censoring is crucial to get unbiased estimates —> hazard models are necessary
right censoring: event that takes place after the observation period ( e.g. observation right–censored customer will never purchase the product or churn but only in the given observation, thus it might happen in other periods)
What do hazard models calculate?
calculate the probability of purchase during certain time intervalls
What do Integrated Choice Models do?
linking brand choice decision to other decisions made at the same time
What is meant with the steady state in Markvo Chain Models?
One of the propterties of Markov chains
–>the *long-term prediction becomes stable after multiple periods –>”steady state”
Steady state is independent of the starting probabilities, only depends on the transition probabilities