phoenix
phoenixPhoenix - Your AI Assistant
Inactive session will be ended in 24 hours
End Session

Algorithm creating for bespoke reviews website

  • Posted:
  • Proposals: 12
  • Remote
  • #2973821
  • Awarded
Augusto P.Sarim I.Parth D.KetriTSOLUTIONS + 7 others have already sent a proposal.
  • 2

Description

Experience Level: Expert
I am looking to create a ranking algorithm for a bespoke review website in a healthcare niche.

The ranking system is to be weighted for 3 factors:

1. The overall review score (maximum is 5 stars)
2. The number of reviews (unlimited number, existing data ranges between a handful of reviews up to 150ish).
3. The number of reviews over time (reviews left in the last month should count as a higher value than those left 12 months ago as a company could have changed it's performance dramatically in a year)

The ranking system will push those with a high score and a high number of reviews to the top. I want to avoid having a company with 5x 5 star reviews ranking above a company that has 30x 4.8 star reviews.

Example:

Company A - 5 Stars - 8 reviews
Company B - 4.89 Stars - 37 reviews
Company C - 4.87 stars - 143 reviews
Company D - 4.50 stars - 85 reviews
Company E - 4.00 stars - 99 reviews

They should look to be ranked in the following order:
C
B
D
E
A

In this example the overall star rating is the most important factor and company C has a high star rating but also a lot of reviews. Company B has a slightly higher star rating, but not as many reviews which is why it comes second. Company D has more reviews but not the rating is now as high as company B which is why it sits underneath. Company A is at the bottom as a few bad reviews could pull it's overall score right down.

Would be happy to discuss the project in more detail with those that feel this is achievable.

New Proposal

Login to your account and send a proposal now to get this project.

Log in

Clarification Board Ask a Question

  • John M.

    Hi Ian,

    Can you please clarify the weighting factors. The first weighting factor is the overall score, which suggests that each company will have just a single score. But the third factor is the score over time, which suggests that each company will have not one score but a sequence of scores. Can you provide a simple example of the time-based data. Would it be of this form (my numbers are plucked out of thin air):
    Company C: (4.87, 143, 1), (4.91, 132, 2), (4.61, 172, 3), ...
    where the third number in each set indicates how many months back the score relates to.

    Someone has asked about the technology stack, but I assume you just want a clear description of the algorithm (perhaps implemented for illustration in a simple way in Excel, Python or similar), which you will then implement yourself. Do let me know if this is correct.

    Regards

    John

  • Callum L.

    Is there any tech stack that you want the logic to be built in? What stack will the server-side logic be built in?