An attempt in Predicting Genre of Music with Neural Networks

Inspiration

My passion for music

What it does

Predicts genre of any Spotify track (WARNING: with only 50% accuracy)

How I built it

35-35-10-10 Multilayer Perceptron; Activation Function - Sigmoid; Optimizer - Adam; Loss Function/Cost Function - Sparse Categorical Cross-Entropy; Metric - sparse_categorical_accuracy Learning Rate - 0.001 Batch Size - 16 Epochs - 15

Challenges I ran into

• Lack of accuracy (duh)
• Missing/Unknown data values
• Understanding Keras Code (this is my first time using Keras/Tensorflow in a project)
• Figuring out Spotify API works
• Uploading the data from Kaggle

Accomplishments that I'm proud of

• This is my first machine learning model (yay)

What I learned

• The importance of mutually exclusive classes in multi-class classification.

What's next for (An attempt in) Predicting Genre of Music with Neural Networks

• Maybe K-means clustering?

Built With

• keras
• opendatasets
• pandas
• python
• scikit-learn
• spotipy
• tensorflow