What did you try?: tried to build a recommendation model with multiple optional custom weighted attributes
What happened?: could not figure out how to apply custom attributes with weight
What did you expect?: build a model with custom weighted attributes
Use case
I would like to develop a recommendation system for cars (yes, vehicles), where user can specify what they are interested in most, less important and least important.
The importance would be described as:
High: the weight value is 10
Medium: the weight value is 6
Low: the weight value is 3
And for every attribute of the car, a user is able specify the importance to them. Which means every attribute is optional. For those attributes that are not specified by the user, it would have default weight at Medium.
For example, a user specifies importance as below:
price: high (more expensive, lower the score would be)
horse power: high (bigger horse power, higher the score would be)
seats: high (the system consider the number of seats as: the more the seats, the higher the score would be)
4WD: low (for those cars that are FWD/4WD, it has higher score)
tank/fuel range: medium (longer the range, higher the score)
etc.
etc.
Now, to build such model, I would expect somehow we could provide user-specified weighted attributes into the model so the result would be user based.
System information
Issue
Use case
I would like to develop a recommendation system for cars (yes, vehicles), where user can specify what they are interested in most, less important and least important.
The importance would be described as:
And for every attribute of the car, a user is able specify the importance to them. Which means every attribute is optional. For those attributes that are not specified by the user, it would have default weight at
Medium
.For example, a user specifies importance as below:
Now, to build such model, I would expect somehow we could provide user-specified weighted attributes into the model so the result would be user based.
Not really sure how to do so. Any thoughts?
Yours, Wilson