Question

1) Assuming that data mining approaches are to be used in the following cases, identify whether the

task required is supervised or unsupervised learning:

a. A service that has a large number of people repeatedly label photos as to whether or

not they contain an object of interest, then wants to use that data to label new photos.

b. A photo-sorting app that takes all your photos of people and automatically divides them

into different groups, each group ideally containing photos that contain a single person.

c. An IT team that wants to identify a network data packet as dangerous (e.g., virus or

hacker attack) based on comparison to other packets whose threat status is known.

d. Predicting whether a start-up company will succeed based on comparing its financial

data to those of similar past start-ups that both succeeded and did not.

e. A real-estate app that seeks to predict the sales price of a home in a certain

neighborhood, given information on recent sales in the area.

f. A grocery store that wants to use sales data regarding items bought together by its

customer base to offer coupons to customers on items they might like to try.

g. When biologists started studying living creatures and were interested in establishing a

taxonomy that would group living things according to various natural similarities and

differences.

h. A news filtering service that seeks to find news articles that contain similar topics.

i. Identifying children with similar learning styles in a classroom so that targeted learning

materials can be used for the different groups.

j.Political candidates attempting to identify groups of voters with similar demographic

characteristics to prepare platform statements to address each group’s unique

concerns.

2) In the game Scrabble, a player has a tray of 7 letter tiles. Consider the word 7-letter word SCIENCE

and answer the following questions.

a. Calculate the probability of randomly selecting each unique letter from this word.

b. Using those probabilities, calculate the entropy of this set of letters.

c. What would be the reduction in entropy (i.e., the information gain), if you split these

letters into two sets, one containing the vowels and one containing the consonants?

d. What is the maximum possible entropy in a set of 7 Scrabble tiles?

e. In general, which is preferable when you are playing Scrabble: a set of letters with high

entropy, or a set of letters with low entropy?

Answer 1

1)

In Supervised Learning, both input variables and output variables will be given.

In Unsupervised Learning, only input variables will be given.

a. A service that has a large number of people repeatedly label photos as to whether or not they contain an object of interest, then wants to use that data to label new photos.

output variable - whether or not they contain an object of interest,

It is Supervised Learning as output variable is defined.

b. A photo-sorting app that takes all your photos of people and automatically divides them into different groups, each group ideally containing photos that contain a single person.

output variable - whether or not the photos that contain a single person.

It is Supervised Learning as output variable is defined.

c. An IT team that wants to identify a network data packet as dangerous (e.g., virus or hacker attack) based on comparison to other packets whose threat status is known.

output variable - whether or not network data packet is dangerous

It is Supervised Learning as output variable is defined.

d. Predicting whether a start-up company will succeed based on comparing its financial data to those of similar past start-ups that both succeeded and did not.

output variable - whether or not a start-up company will succeed

It is Supervised Learning as output variable is defined.

e. A real-estate app that seeks to predict the sales price of a home in a certain neighborhood, given information on recent sales in the area.

output variable - sales price of a home

It is Supervised Learning as output variable is defined.

f. A grocery store that wants to use sales data regarding items bought together by its customer base to offer coupons to customers on items they might like to try.

No output variable - We have not given criteria to determine which coupons to offer.

It is Unsupervised Learning as output variable is defined.

g. When biologists started studying living creatures and were interested in establishing a taxonomy that would group living things according to various natural similarities and differences.

No output variable. We have not given by which natural similarities and differences we need to group living creatures.

It is Unsupervised Learning as output variable is defined.

h. A news filtering service that seeks to find news articles that contain similar topics.

No output variable. We have not given criteria to determine the similarity and differences of topics.

It is Unsupervised Learning as output variable is defined.

i. Identifying children with similar learning styles in a classroom so that targeted learning materials can be used for the different groups.

No output variable. We have not given criteria to determine the similarity and differences of learning styles.

It is Unsupervised Learning as output variable is defined.

j.Political candidates attempting to identify groups of voters with similar demographic characteristics to prepare platform statements to address each group’s unique concerns.

No output variable. We have not given criteria to determine the similarity and differences of demographic characteristics.

It is Unsupervised Learning as output variable is defined.