Boosting Model Accuracy: Techniques I Learned During My Machine Learning Thesis at Spotify (+Code Snippets)
<p>In 2021, I spent 8 months building a predictive model to measure <em>user satisfaction</em> as part of my Thesis at Spotify.</p>
<p><img alt="" src="https://miro.medium.com/v2/resize:fit:700/1*SUiVM45BO_U51aRFrOJ1pw.jpeg" style="height:549px; width:700px" /></p>
<p>Image by Author</p>
<p>My goal was to understand what made users satisfied with their music experience. To do so, I built a LightGBM classifier whose output was a binary response:<br />
<em>y = 1 → the user is seemingly satisfied<br />
y = 0 → not so much</em></p>
<p>Predicting human satisfaction is a challenge because humans are by definition unsatisfied. Even a machine isn’t so fit to decipher the mysteries of the human psyche. So naturally my model was as confused as one can be.</p>
<h2>From Human Predictor to Fortune Teller</h2>
<p>My accuracy score was around 0.5, which is the worst possible outcome you can get on a classifier. It means the algorithm has a 50% chance of predicting yes or no, and that’s as random as a human guess.</p>
<p><strong>Website</strong></p>