Looking for a ranking method to produce a list of products from best to worst given some features

38 Views Asked by At

I am pretty new to data mining and I am trying to help a friend of mine who is working on his graduate thesis in economics. The main purpose of his work is to rank a number of items (smart home devices) given their features and using review scores from the users. To be more clear I will give you an example of what we are trying to do... He collected info about smart home devices ( as for example light bulbs) that are: number of communication protocols, number of items with which they can communicate, number of apps available, average rating from the users, number of reviews on the web and so on... My question is which approach could we use to rank these products? neural networks? I am looking for something that is not really difficult to implement. Hope someone could help me or suggest me something to look for... Thank you for your help!

1

There are 1 best solutions below

1
On

Probably the easiest thing to do (e.g. what US News and World Report does to rank colleges) is to divide each feature by its standard deviation and then assign a numeric weight to each feature and then rank items according to their weighted sum score. If you want something more agnostic to the weights, you can sample positive weights in an angle independent fashion by using independent samples from a Gaussian distribution for each weight, and then compute the average of all rankings for all your sampled weights and then do an overall ranking based on sorting the average rankings.