Overview

To make a matching algorithm to pair up accountability partners. It ingests data points and outputs a list of matches (users).

User: One who has enrolled in the program and is waiting for matches.

 

How the matching algorithm works

Find similar people:

• Actively filters the sample space based on gender preference (M, F, Any)
• Groups people into 2 categories:
  • C1: Age bracket (+-) 5 years [Value 5 may be changed for edge cases]
  • C2: Other people
• Score each category based on passion_vector similarity
• For C1, express passion similarity out of 100 and scale it by 3/2 for people with occupation match
• For C2, express passion similarity out of 100 and allow only those with an occupation match and halve score for others
• Sort all users based on score

Find different people:

• Actively filters the sample space based on gender preference (M, F, Any)
• Groups people into 2 categories:
  • C1: Age bracket (+-) 5 years [Value 5 may be changed for edge cases]
  • C2: Other people
• Score each category based on 100 minus passion_vector similarity
• For C1, express passion similarity out of 100 and scale it by 3/2 for people with occupation match
• For C2, express passion similarity out of 100 and allow only those with an occupation match and halve score for others
• Sort all users based on score