Please enable JavaScript.
Coggle requires JavaScript to display documents.
Chapter 10.2/3 Transforms (Different Types of Transformations (Addition +,…
Chapter 10.2/3 Transforms
used to create new features
comparison of 2 columns into 1
additional predictive ability
Different Types of Transformations
Addition +
ex: number of children +1 = family size
predictive signals increased
Subtraction -
similarity/differences in columns more apparent
Preferred Inside temp-current outside temp
'+ like warmer temps
'- like colder temps
close to 0- prefer outside activities
Absolute abs()
measures actual distance b/w 2 numbers
Multiplication *
measure interaction effect/moderated relationship
customer service experience * how good or bad=likelihood of churn
Division /
show hidden information some algorithms can't see
income/ number of children= allocation of funds per child
Less than <
number of seats in a car<number of family members
help predict new car purchased, based on size
Less than or Equal To <=
number of bedrooms<=family size
predict purchase of bunk beds
Greater Than >
family size>number of seats in car
less likely to take road trips
Greater than or equal to >=
number of seats in car>=family size
less likely to buy a minivan
Not Equal !=
when two data points are not the same
predict that something is wrong/inconsistent
Equal ==
If data points are the same, may cancel each other out or indicate increased likelihood of target phenomena occurring
Exponential **
show exponential relationships
P=Ce^(rt) is read as C
e**(r
t) (to the)
applied 1 column at a time
Natural Log Log()
linearize exponential data
Square Root Sqrt()
similar to log
Square square()
large values even larger
square(standard deviation)= variance