Distance & Scaling

Question 1

Euclidean Measure

Answer 1

square root of (x2-x1)^2 + (y2-y1)^2

Question 2

Manhattan Distance

Answer 2

Sum of the absolute differences between the coordinates of two points

Question 3

Minkowski Distance is

Answer 3

A generalization

Question 4

Minkowski Distance r = 1

Answer 4

Manhattan

Question 5

Minkowski Distance r = 2

Answer 5

Euclidean

Question 6

Minkowski Distance r = inf

Answer 6

Supremum Distance

Question 7

As r is higher in minkowski distance

Answer 7

Gives more weight to larger differences, emphasizing outliers

Question 8

Mahalanobis Distance

Answer 8

Uses variance to plot how close two points are

Question 9

Standardization

Answer 9

Transforms data with mean=0 and std. dev = 1 (z-score_

Question 10

Normalization

Answer 10

Scales a variable to have value between 0 and 1

Question 11

When do you want to standardize distances

Answer 11

If the scales differ significantly

Question 12

One Hot Encoding

Answer 12

Technique to convert categorial data into a binary matrix (numerical representation)

Question 13

What happens if we dont scale

Answer 13

Some algorithms will be slow in converging
Features with high magnitudes will dominate the distance calculations

Question 14

In one hot encoding, each category is represented as

Answer 14

a binary vector with a single 1 and all other positions as 0

Question 15

When we scale, do we lose geometric representation of each point with respect to its neighbors?

Answer 15

No, the points stay exactly the same

Question 16

Should we scale with decision trees?

Answer 16

No, don’t need to scale and we lose the original value of the predictor which would make decision less explainable

Question 17

Scale then split or split then scale?

Answer 17

Split then scale. So that you don’t leak information from the training set into the test set

Question 18

Discretization

Answer 18

Sort numbers, create split points, map split values to discrete categorical variables

Question 19

Standardization is essentially

Answer 19

Feature Scaling

Question 20

Standardization formula

Answer 20

replace values with z score, mean of 0 and std dev of 1

Question 21

Mean Normalization

Answer 21

Redistributes with range [-1, 1] mean =0

Question 22

Min-max scaling

Answer 22

Redistributes with range [0,1]