- Nathaniel Haddad haddad.na@northeastern.edu
- Winston Moh Tangongho mohtangongho.w@northeastern.edu
- Northeastern University
- Disclosure: this is an academic project
Our objective for this project is to provide YouTube content creators with a detailed trend analysis and probabilistic interpretation of what makes a top trending video on the platform in order to maximize the reach and quality of creator content. To do so, we performed a number of small experiments including learning word embeddings and examining linear relationships in comments, titles, and descriptions in YouTube videos using word2vec neural networks. A second experiment uses a VGG-19 neural network in a self-organizing map as part of a detailed analysis of video thumbnails. We also performed a number of other small tasks, and learned a lot along the way!
- clone the repository above
- install the dependencies
- launch Jupyter notebook
- Run the notebook
[1] Dong, Yuxiao, et al. “Will This Paper Increase Your h-index? Scientific Impact Prediction” Johnson, Reid A., Chawla, Nitesh V., Proc. of the 8th ACM International Conference on Web Search and Data Mining (WSDM'15) https://arxiv.org/abs/1412.4754
[2] CS6140 Machine Learning Class Resources