Slideshow by Mario Marquez

mariomarquez4 — 2025-05-12 19:24:31 UTC

It is a computer system that can adapt and learn using algorithms and statistics to analyze and make a pattern in the data, but does not use instructions.

mariomarquez4 — 2025-05-12 20:07:32 UTC

Filtering methods find the best information in a data set that helps make better predictions.

mariomarquez4 — 2025-05-14 20:03:04 UTC

Recommendations that suggest similar items that other people have looked at that you might find interesting, but are still similar to the items you like.

mariomarquez4 — 2025-05-14 20:07:42 UTC

A recommendation that predicts users' preferences and is based on the preferences of similar users. Finding similar users who have similar interests and then recommending them to another person who has similar tastes.

flow chart

mariomarquez4 — 2025-05-14 20:09:24 UTC

mariomarquez4 — 2025-05-14 20:26:33 UTC

The purpose of listening to all of the 23 songs and rating them was so that we could test and try each machine learning model to see which machine learning made the best prediction.

mariomarquez4 — 2025-05-14 20:32:58 UTC

The 3 attributes I chose were speechiness, Danceability, and loudness. How I chose them, which attributes I chose, and where they are closest together and matched well. Why I chose them was because they appealed to me, and they also go well together.

mariomarquez4 — 2025-05-14 20:49:12 UTC

We chose the appropriate complexity level by the actual graphs by hand this helped us pick which points on our graph is the best match. We also did this by looking at which graph had the least errors the graph with the least amount of errors was the most appropriate complexity level.

one attribute

mariomarquez4 — 2025-05-16 16:07:52 UTC

about one attribute module

mariomarquez4 — 2025-05-16 16:12:33 UTC

This attribute model predicted how I would rate this song as good. Because in this model, even though the line is far off the numbers, there is little difference in the error between training and test. The prediction was not far off what I would rate it, the prediction was 5.77.

mariomarquez4 — 2025-05-16 16:18:58 UTC

Attribute module 2

mariomarquez4 — 2025-05-16 16:23:19 UTC

Attribute 2 was also accurate in the rating. The prediction that it gave me was 4.9, which is not a bad rating for the song.

class matrix

mariomarquez4 — 2025-05-16 16:27:02 UTC

recommending system

mariomarquez4 — 2025-05-16 16:30:04 UTC

mariomarquez4 — 2025-05-16 16:36:07 UTC

My results for my ratings are good; out of the 8 predictions, 6 of them are not far off from my actual rating.
I didn't have any real surprises.
This adds to my understanding that collaborative filtering is more accurate than content-based filtering in some situations.
I feel like there should be more of a variety of attributes and more main stream songs or older classic because each song apeals to people different and if you choose a bunch of songs a person doesn't know or doesn't like then it can lead to lop sided data.

Thoughts on Ethics of machine learning.

mariomarquez4 — 2025-05-16 16:40:13 UTC

I feel Machine learning is good in some ways because if you are answering questions or rating something, it can eliminate some type of bias. The reason for this is that if you are in a group but can't share ratings, it eliminates the bias of that person trying to pick what their friend/ other people.

thoughts on info about machine learning

mariomarquez4 — 2025-05-16 16:47:13 UTC

I feel like people should know that machine learning is always growing. The reason is that almost every person in the world uses the internet to look up how to cook something or how to fix something. In this, computer programs can pick this up and suggest these types of things to you in the future. This could mean something as big as Google can pick something up, like how to fix a car problem from a video a person posts. That is just one of the ways machine learning grows bigger and bigger.

biases

mariomarquez4 — 2025-05-16 17:04:32 UTC

I feel like both methods of machine learning are selection bias because I feel like the data is based on what that person likes, but also with collaborative filtering, the data can be a bit off because people can go off what other people like.