Unit 7 Project by Jason Gorovyy

Machine Learning and Filtering Methods

jasongorovyy — 2025-05-16 00:54:44 UTC

Machine learning is when the program takes previous data from patterns and analyzing and makes predictions. Filtering Methods are ways that ai and computers can filter out things so we see what we want. Content-Based Filtering is when programs recommend certain items based on previous buys and searches you have made. Collabrative Filtering is when a program recommends items and different thing by basing it off of other people with similar interests.

Collecting Data

jasongorovyy — 2025-05-16 09:33:59 UTC

We recorded our ratings of the 23 songs because it helped diversify our ratings and certain attributes for the machine learning to help us predict what our ratings will be for a song we haven't heard yet. My top 3 attributes are Tempo, Popularity, and Speechiness because that is what the best fit for my song ratings where.

Filtering Models

jasongorovyy — 2025-05-20 18:33:16 UTC

The predictions weren't the best but most of them were only a couple of numbers off, so I think it worked decently good.

Filtering Models 2

jasongorovyy — 2025-05-20 18:35:45 UTC

These ratings were very inconsistent, they were either really close or very far off. I think the first set of ratings was better because it was more consistent.

Filtering Models 3

jasongorovyy — 2025-05-20 18:41:28 UTC

The two highest rated songs that I rated similar to other people were Jasmine and Trevor, and people who rated similar to the lowest rating were Lily and Milena.

Filtering Models 4

jasongorovyy — 2025-05-20 18:43:35 UTC

The results were very far off of my actual results. I was surprised that the final predictions were the worst compared to the first two sets of predictions. If I rated more accurately I think the results would have been better.

Data Ethics of Machine Learning

jasongorovyy — 2025-05-20 18:46:59 UTC

I think that Machine Learning could be a very useful tool for some people and if you need to make predictions on things that don't have to be 100% accurate Machine Learning could be a very good tool to use. The general publ8ic should start using Mcahine Learning because it could keep track of the patterns of the data they need.