Predict the popularity of videos at the time of upload
Currently reworking solution. CollectData.py contains code for a more methodical approach to sampling videos and collecting data. This approach to sampling introduces less bias and provides a better representation of the YouTube network than my previous approach. Still working on storing only processed audio data, rather than entire wav file (wav files tend to be very large). Dataset
This entire pipeline (data collection, processing, classification) is compressed into Pipeline.py (currently undocumented). Dataset