Lecture 2 Flashcards
What can you use Big Data for?
Counting (descriptive research)
Prediction (association)
Approximating experiments (causal)
Why we need Web analytics
to understand what are visitors doing to our website
Retention
visitors who Returned to the web page
How to calculate retention?
visitors who returned/ total visitors
Bounce rate
( come-see-didn’t like it- left the page)
people who visit the page without any clicks
How do calculate the bounce rate?
amount of visits without any click/ total amount of visits
How do calculate the conversion rate?
the amount of visits with purchases/ total visits
What are the steps of the “Path to purchase”
Awareness -> Beliefs -> Attitude -> Behavior
What is the co-occurrence in the Big Data approach
the more often brands appear together, the more competitive they are
Lift
the ratio of the actual co-occurrence to the occurrence we would expect if brands were independent
How to calculate the lift?
Lift = P(A,B)/P(A)P(B)
Lift > 1
brands are close competitors
Lift < 1
The brands are not close competitors
How does Matching approximate an experiments ?
finds units that are alike in terms of characteristics, so that any change in the outcome is likely due to the treatment
Ex. The Twins
Mr. Treatment vs Mr. Control
What is the reason if there is a difference in the outcomes after doing a matching?
it is due to a difference in treatment assignemnt
