Similarity
- Lets have three existing cases:
Name | Income | Expenses | Assets | Result |
Chris | 25000 | 20000 | 50000 | Yes |
Pat | 25000 | 25000 | 50000 | No |
Sandy | 50000 | 45000 | -25000 | No |
- Mel wants to take out a loan (everyone asks for 10000), and makes
35000 per year spend 25000 and has 10000 in assets. Should Mel
get a loan.
- You need to compare to the others. How do you compare? Which is
Mel closest to.
- Mel shares one feature exactly with Pat, so is that closest?
- Measuring similarity is really important.
- Here you might have a complex way of comparing, get total surplus
multiply that by 5, and add to assests.
- So Mel is 60000, Chris is 75000, Pat 50000 and Sandy 0
- Which is Mel nearest to. Here you could say Pat(No), but you could
also say Chris(Yes) or Maybe because it's between a No and Yes
- In this case we've taken all the features and combined them to
get one value. You can't always do this and it's not always wise.