SDS 433: Data Science Trends for 2021 — with Ben Taylor

Framework Battle

The debate between TensorFlow and PyTorch continues to be a hot topic in the data science community. Jon Krohn shares his experiences with both frameworks, noting that PyTorch offers a more intuitive and Pythonic experience compared to TensorFlow, especially for those familiar with libraries like NumPy and SciPy 1. Ben Taylor reminisces about the simplicity of Keras before its integration with TensorFlow, highlighting its ease of use and beautifully written code 2. Jon expresses his preference for PyTorch, even predicting its rise in popularity over TensorFlow due to its user-friendly nature 3.

   

New Tools

Emerging machine learning packages are set to revolutionize the industry in 2021. Ben Taylor is keen on exploring these new tools, which promise to enhance the capabilities of data scientists 4. The shift to remote work has also opened up new opportunities, making it easier for data scientists to work from anywhere, as long as they have a reliable internet connection 5. Ben reflects on the past challenges of securing remote work in data science, which have now become a thing of the past due to the pandemic's impact on work culture.

   

AutoML Leadership

DataRobot is at the forefront of AutoML technologies, offering end-to-end solutions that streamline the machine learning process. Ben Taylor discusses DataRobot's journey, highlighting its numerous acquisitions and its leadership in applied AI across various industries 6. He emphasizes the importance of ML productionization, which involves deploying models efficiently and adapting to changes like new data or unforeseen events 7. Ben humorously shares his past experiences with model retraining, underscoring the need for a robust process to handle continuous learning and deployment.