Question

Given the training data in Question 1 below| (on buying RRSP8), predict the class of the following new example using k-nearest-neighbor classification fork = 5: sector = oil industry, income = medium, self-employed = yes, credit-rating fair. For distance measure, use the following similarity measure: similarity(tupleAtupleB)-4-.(w"S(ab/4), where S(ab) is 1 if parameter a equals parameter b and o otherwise The parameters atand biare either Sector, income, self-employed, or credit-rating. The weights wiare all 1, except for income, which is 2. Please show your calculations or explain the steps. Note: the similarity measure is the opposite of the distance measure. Sector of activity Self-Employed Credit-rating Person ID Class: Buys-RRSP Income farming medium fair 1 no no farming fair low 2 yes no oil industry medium fair no no oil industry low fair yes 4 yes oil industry medium excellent 5 yes yes banking medium excellent yes no oil industry high fair 7 no no oil industry high excellent no no banking high fair yes no farming low excellent 10 yes yes banking low excellent 11 yes yes farming medium fair 12 yes no high banking fair 13 yes yes farming medium excellent 14 no yes

Answer 1

Solution

And We have to predict., sector = oil industry income = medium. self-employed = yes, credit rating fair

Also the similarity measure is given as : similarity (tuple A, tuple B) =

Wi (S(ai, b)) i=1

where s(a, b)= 1 if parameter "a" equals parameter "b".. and =0 otherwise The parameters a(i), b(i).. are ether Sector , income, self-employed, or credit-rating. The weights w(i) are all 1. except for income. which is 2.

NOTE: For calculating similarity measure we just use the given formula and corresponding value of s and W. depend, on the case if they match our training set and the query or not

Person Sector of Activity Self Income Credit Similarity employed Rating ID Measure Value 0+2 0+1 3 1 farming medium No Fair 0+0+1+1 2 2 farming low Fair yes oil industry 3 medium No Fair 1+2+0+1 4 1+0+1+1 3 oil industry low fair yes oil industry medium excellent 1-2+1+0 4 yes 0+2+0+0 2 6 banking medium excellent no 1+0+0+1 2 oil industry 7 high fair no oil industry 8 high excellent 1+0+0+0 1 no banking high fair 0+0+0+1 1 no 0+0+1+0 1 10 farming low excellent yes excellent 0+0+1+0 1 11 banking low yes farming 0+2+1+1 1 12 medium fair yes 0+0+1+1 2 13 banking high fair yes 0+2 0+0 2 14 farming medium excellent no O st LO

Person Sector Credit Similarity Rank Income Self- Can we Y=Category ID of Rating include it Measure Highest Of Nearest employed in (K 5) Activity Similarity Nearest Value Neighbour Neighbour Measure farming medium 0+2+0+1 1 No Fair 3 YES No = 3 Fair No 2 farming low 0+0+1+1 5 YES yes = 2 3 oil medium No Fair 1+2+0+1 1 YES No = 4 industry 1+0+1+1 4 oil low fair 4 YES Yes yes industry = 3 excellent 1+2+1+0 5 oil medium YES Yes yes industry =4 banking medium 0+2+0+0 excellent 6 NO Yes no = 2 fair 7 oil high 1+0+0+1 NO No no industry = 2 No oil high excellent 1+0+0+0 10 NO no industry = 1 banking high 0+0+0+1 9 fair 11 NO Yes no = 1 LO CO st LO

excellent banking low 0+0+1+0 13 NO Yes yes = 1 12 farming medium fair 0+2+1 1 14 NO No yes = 1 8 13 banking high fair 0+0+1+1 NO Yes yes = 2 farming 14 medium 0+2+0+0 NO Yes excellent no = 2 11

So we can see that we have 3 No and 2 yes so our final answer class label will be "No".

(Note: here we had to choose one among several choices for the fifth neighbour so we selected one randomly. and there will be a change of answer if we had chose another one. So in such case we just select a random one and make our decision of the class)