Week 3

Question 1

Extraneous information can do what to a model

Answer 1

complicates the model
harder to correctly interpret the solution

Question 2

Outlier

Answer 2

data point very different than the rest

Question 3

contextual outlier

Answer 3

outier that relies on the context of the points

Question 4

collective outlier

Answer 4

something is missing in a range of points

Question 5

Plot for finding ourliers

Answer 5

box and whisker plot

Question 6

Box and whisker plot

Answer 6

Top and bottom are the 25 and 75th percentile of the values and horizontal line is the 50th percentile. The vertical line up and down are called the whiskers. Points outside range are outliers

Question 7

CUSUM Equation

Answer 7

St = max{0, S_t- (S_t - mu - C) }

If S_t > T

Question 8

WHen deciding what to do with outliers,

Answer 8

it is important to understand how they ended up in your dataset. Is it an error or a real data point. If it is an error, you may want to remove the data point

Question 9

What things do you need to look at when assessing an outlier

Answer 9

1) Where did the data come from
2) How was the data compiled?
3) The situation in which the data was compiled

Question 10

CUSUM from high level

Answer 10

Determines whether the mean of the observed distribution has gone gone beyond a critical level. It can detect when a process gets to a higher level than before, a lower level, or both