Decision making tool: Lessons to have a best choice.

1. Planning:

You're going to buy a new Macbook. Of course, it is a hypothesis ;)

There are a wide variety of choices at FPTShop (the most popular tech shop in Vietnam): 

Macbook Air 13 128GB MQD32SA/A (2017); 

Macbook Air 13 256GB MQD42SA/A (2017); 

Macbook Pro 13 inch 128GB (2017); 

Macbook Pro 13 inch 256GB (2017); 

Macbook Pro 13 inch Touch Bar 256GB (2017); 

Macbook Pro 13 inch Touch Bar 512GB (2017); 

Macbook Pro 15 inch Touch Bar 256GB (2017); 

Macbook Pro 15 inch Touch Bar 512GB (2017); 

Macbook 12 256GB (2017); 

Macbook 12 512GB (2017); 

Macbook Pro 13 256GB (2016); 

Macbook Pro 13 Touch Bar 512GB (2016). 

12 cases, so what's the best one for you?

2. Objectives:

Twelve MacBooks are evaluated based on six criteria. The criteria will depend on your preferences.

• price (to be minimized)

• screen (to be minimized)

• RAM (GB, to be maximized)

• CPU (GHz, to be maximized)

• weight (Kg, to be minimized)

• battery (Whr/cells, to be maximized)

• VGA (do not care)

After that, there are 2 main stages as following:

Stage 1: Proposed parameters such as weights and thresholds will reflect your own opinions. Stage 2: Utilize outranking methods to select the best MacBook or, depending on the methods, rank all the models by using MATLAB (or Scilab) to program the Electre and Promethee methods.

3. Processing: 

i. Proposing the parameters (weights, threshold, etc.)

The project proposed the following parameters: weights, thresholds, and other relevant details.

Weighted criteria: The relative importance of each criterion is quantified by a weight. In my opinion, I prefer a MacBook that is more affordable but also faster. Therefore, I assign a weight of 0.4 to price (the highest importance), while the weights for RAM and CPU are 0.15 each, and all other criteria are weighted at 0.1.

Preference threshold: This threshold defines an interval where the difference between two alternatives is considered insignificant. This means that a decision-maker can freely choose between any two alternatives marked as indifferent. Therefore, I suggest using this parameter as shown in the table above. Veto threshold: This is a limit beyond which the credibility of the outranking relation between two alternatives is denied. Therefore, I also propose them as the basis for the table.

2. Outranking methods to select the best Mac

2.1 Promethee I

This is the Matlab code that was created to compute the multicriteria preference degree table (see pref.m):

function conc=preference(data,minmax,weight) number_actions=size(data,1); D=data.*(ones(number_actions,1)*minmax); conc=zeros(number_actions);

for i=1:number_actions-1

for j=i+1:number_actions conc(i,j)=sum(weight.*(D(i,:)< D(j,:))); conc(j,i)=sum(weight.*(D(j,:)< D(i,:))); end end

Then, I enter and display the data: 

TBC ... If you're interested in the algorithms, I can send you the full version. It's quite similar, but my public version has differences.