The @taylorswift13 #datascience Competition vs @jedludlow
??Jepson Taylor
CEO/founder VEOX Inc | Fasted Ultra Runner | Contributing Faculty AIMasterClass NYU | Keynote Speaker | former child
When it comes to prediction @jedludlow doesn't mess around. Jed came out swinging on day one with a 99.37% accuracy on the private validation set! For a relative comparison Google's prediction engine is only pulling in 94.86%.
The original competition was posted here and is attempting to predict if a sub second clip of audio will be a viral success for Taylor Swift.
The future applications are fun, go get your $1M seed for a new big data application.
Have you ever heard a cord or chorus where you thought "Now that is good...". Well, with enough data and the latest machine learning algorithms couldn't a computer do that for you? Imagine using big data to consult artists in realtime in the studio as their iterate on their creations. 98% confidence that this is a killer chorus, but your intro needs some work...
There are less than 2 days left for the competition to close. To help you I have included a full working code solution with the data here:
https://taylorswiftdata.s3-website-us-east-1.amazonaws.com/data_comp.html
With that you should be able to get a solution uploaded and scored in <2 minutes. An online kaggle-style platform was stood up for realtime scoring and peer ranking here
I Still Don't Care
Why should I care? What is in it for me? 2 things.
(1) Visibility/Marketability:
Employers love this stuff. They watch you fight and use results to compare who is epic and who is boring. Win enough data competitions and employer's pupils will dilate when they look at you instead of squint as they determine where to place you in the pecking order of applicants.
(2) You Will LEARN Something
The top contestants are doing things Google's ensemble prediction API wasn't. They are all beating the traditional random forest by a large margin. If your submission is higher than the default code you will be in the know. I will make sure you are. You will learn from the best in the game and learn new skills you can take to your next problem.
Competition ends this Friday at midnight.
Keywords: Random Forest, Deep Learning, Data Competition, Big Data, Predictive Modeling, Big Data Analytics, Big Data Analysis.
CDO | CIO | CTO | Founder | Entrepreneur | Emerging Technology Strategist | Board Advisor | Building Innovation Leaders One Project At A Time
10 å¹´Proof once again that the real data gurus are located right here on the Silicon Slopes...
Lead Data Scientist at Enova International
10 å¹´"Employers love this stuff." Quoted for emphasis.
I build and lead high-performing teams through curiosity, collaboration, and trust. Together we use data, AI and ML to make better decisions and improve business outcomes.
10 å¹´Have to say, its kind of depressing to hit 97.8 percent accuracy and still feel like I'm overmatched. Nice job though, Jed! I'm interested to hear how you tackled this after the competition ends.
Chief Technologist | High-Accuracy Inspection Systems | Energy Pipelines
10 å¹´Thanks, Ben. I guess the cat's out of the bag now!