Unit 7 Project by Tatiana Cerezo

Content Based Filtering Methods

tatianacerezo — 2025-05-14 19:57:48 UTC

With only using one attribute model, it was predicted I would rate the song a 4.66 when in reality I would have rated it a 3 or 4. So, all in all, the prediction was pretty accurate.

Content Based Filtering Methods

tatianacerezo — 2025-05-14 19:58:06 UTC

With the two-attribute model, it was predicted that I would give the song a 6.1 when I would have given it a 4. The prediction was inaccurate because I would have given it a lower rating.

tatianacerezo — 2025-05-14 19:58:19 UTC

Only four of my results were somewhat similar to my actual ratings; the rest were completely off. The surprise was how different the predictions were, but it was likely due to the lack of pattern in my choices. This shows that collaborative filtering is not always accurate and can only be effective sometimes. You can change your model to be more accurate by giving more diverse ratings and having a variety of people with different tastes.

S2 Level of Complexity

tatianacerezo — 2025-05-14 19:59:30 UTC

You have to choose the level of complexity that has the lowest error amount and has it so that the two numbers are as close as they can be to each other. Be sure to use your activity to help explain.

S 1 Define Machine Learning

tatianacerezo — 2025-05-14 21:49:43 UTC

Machine Learning is the use and development of computer systems that can learn and adapt without following specific instructions. Examples include spam filtering or fraud detection.

S1 Filtering Methods

tatianacerezo — 2025-05-14 21:50:17 UTC

Content-Based Filtering recommends items to users based on the attributes of items they've previously interacted with. Collaborative Filtering recommends items by leveraging patterns in user-item interactions across a user community. Instead of focusing on item attributes, it identifies users with similar preferences (user-based) or items that are frequently liked by the same users (item-based).

S2 data collection

tatianacerezo — 2025-05-14 21:52:01 UTC

The purpose of listening to all 23 songs was to gain a wide range of data and your song preferences. We assigned ratings by listening to 20 seconds of each song and rating them from 1 - 10 based on how much we liked them. We will input this data into Colab for the machine learning. My top 3 attributes were loudness, danceability, and energy. I chose those three because I prefer more upbeat songs.

S3 Collaborative Filtering

tatianacerezo — 2025-05-14 21:57:35 UTC

The student with the most similar taste in songs based on ratings was Jasmine Hale.

S4 Data Ethics Of Machine Learning

tatianacerezo — 2025-05-14 21:59:14 UTC

One of the most important things for people to know about machine learning is that it is everywhere, and transparency is limited. The biases that can affect machine learning are data bias, selection bias, confirmation bias, measurement bias, and label bias.