Objective: Create a classification model for Spotify Listening Personas

Goals:

• Work with real, up to date data
• Store data in the cloud
• Use relevant libraries
• Choose appropriate algorithms such as KNN, Decision Trees, Random Forest, Naive Bayes and find the “best” model (How do I know what the best model is?)
• perform validation tests on data
• outline assumptions of the model
• Be able to repeat the process on new data and upload it to the cloud (historical data cycle)

Outline:

• [x] Get data from Spotify Charts through API scraping and InfluxDB
• [ ] Clean and prepare the data, feature scaling and regularization for the models
• [ ] Explore and Visualize Data
• [ ] Ask Questions About the Data
• [ ] Use those questions to determine how you will build model, read Data 8 https://inferentialthinking.com/chapters/17/Classification.html
• [ ] Build Model using K-means, K-NN
  1. K-NN: set list of criteria using pandas queries to make a new column (’Mood’) classifying the type of song it is: [Chill, Upbeat, Crazy, Slow, etc]
     1. then use KNN to predict a song’s ‘Mood’ based on the features
     2. K-means: cluster the songs into k (first find optimal k) different listening personas
• [ ] Create A Dashboard for the model using Dash/Plotly and deploy it!
  1. Allow for new data to be read into the web-app and new predictions and metrics to be generated!
  2. Data Pipeline: Data collection → Wrangling → Feed it into the model → Model results via Dashboard/App
• [ ] Optimize Model?
• [ ] Explain the importance of this project: How would this be beneficial to Spotify?