Please enable JavaScript.
Coggle requires JavaScript to display documents.
Ch. 6 Define Prediction Target (Importance to ML (Titanic example (most…
Ch. 6 Define Prediction Target
What Is It?
the behavior of the "'thing" we need to know about in the future
possible to obtain right out of dataset
Lending Club example
want to know how much of a loan person is likely to pay back by the end of loan period
target feature column called"loan_is_bad"
data scientist or SME decided how to define target feature
Importance to ML
allow humans & machines learn how associations drive outcomes
Titanic example
Target: "Survived"
most important relationship to survival is the sex of passenger
the Feature
Kinds of Targets
classification
predicts category where new case belongs
ex: who gets divorced in first 10 years? TRUE/FALSE
examine possible states of variable - can it be simplified?
make sure not to overlook crucial info
regression
predict target numeric values
ex: predict how many years newlywed couple will be married
more complex than classification
can be made into classification if simplified into categories (buckets)