title: "SDS 300 Project Pt. 2" author: "Leah Harries + Lorelei Ing + Destiny Cecchi + Danielle Weiss" format: html

The purpose of Agnostic learning is to reduce the assumptions made about the target function and discover a hypothesis that approximates it with minimal assumptions (Kearns et al., 1992). This approach utilizes dynamic programming and loss functions to find the best approximation. Agnostic learning, unlike other learning techniques which assume a specific fixed function form, allows the model to learn from data with complex relationships, such as those involving high-degree interaction among features. This means that the model can learn to predict categorical outcomes or continuous-valued outputs based on generic features. To understand the concept of agnostic learning, consider the binary classification problem of identifying ripe apples. Let the hypothesis class H contain all possible ways of distinguishing between ripe and unripe fruits, which might include features such as color, size, texture, or sweetness.

title: "SDS 300 Project Pt. 1" author: "Leah Harries + Lorelei Ing + Destiny Cecchi + Danielle Weiss"

Determining wether an email is spam or not