Note: All starter materials will be made available by 21st June.
I’m really excited to invite you all to the Chai Time Data Science, Insights Kaggle competition
CTDS.Show is a podcast by Sanyam Bhutani where he interviews his ML Heroes.
The competition is aimed at articulating insights from the Interviews with ML Heroes. Check the judging criteria for complete details.
We will also be hosting sessions and sharing resources aimed at:
- Introduction to Kaggle
- Data Exploration: EDA Session
- NLP Session
There will be 2 (repeated) sessions on getting started on Kaggle and EDA with active support throughout the run on the forums: Keep an eye out on this thread for details.
The dataset consists of detailed Stats, Transcripts (To be added in the next version) of CTDS.Show.
Any External data usage is allowed as long as it’s shared with other participants (Thread). The usage of video and audio files of the podcast is also allowed, participants are free to download these from YouTube/Audio hosting for usage with this competition.
The goal is to use these and come up with interesting insights or stories based on the 100+ interviews with ML Heroes (CTDS is a continuation of blog interview series having 25 interviews. Note, you’re allowed to use the blog versions for finding insights)
Launch: On the anniversary recording date of CTDS, 21st June, 2020
Session/Workshop Dates: (Weekends/TBD will be updated during launch)
End: 14th July, 2020.
A 150$ Cash Prize pool will be distributed between winners: 75$ to each winner.
Participants will also be awarded a certificate by MLT, DSNet and Made With ML if they submit at least 1 Kernel that is original work, the notebook should ideally talk about a particular theme/story instead of simple EDA and shouldn’t follow a voting ring pattern.
A female and a male winner will be invited for the 1 year anniversary interview episode on the podcast to share their insights. (Can you pinpoint to the date using the dataset? )
- Abhishek Thakur (The only 4x Kaggle Grandmaster in the world) has very kindly agreed to gift 7 Physical copies of his book Approaching (Almost) any ML Problem to all of the winners (2 + 5 categorical). We’d highly recommend checking out his YouTube Channel for many videos along the lines of the book title.
- Manning publications has agreed 4 (2 each) of their books and winners can decide between either of them:
- They have also provided 40% discount coupons valid across all of their materials!
Please use: mtpctds20 coupon once the contest goes live to utilise it.
Note: None of the links above are affiliate links, both Abhishek Thakur and Manning publications were kind enough to support our contest and is just a reflection of their support for the Data science community.
The scoring will be based on points out of 100, based on 5 criteria highlighted below.
Every participant will be evaluated out of 5 points by the 4 judges allowing the maximum score achieved to be 100. (5 points * 5 criteria * 4 judges)
Winners will be chosen based on the highest score achieved overall along with special recognition to participants with highest score achieved in individual criterion.
Final LB will be published on the 1 year anniversary of the Podcast.
- Presentation : How well is the notebook written in terms of code quality, text description and grammar?
- Storytelling : Is there a natural flow of the story that connects various points?
- Visualizations : Are the visualizations appealing, understandable and aligned with description?
- Insights : Are the insights relevant, useful and actionable?
- Innovation : How novel and creative are the ideas and approaches?
We are really grateful to the stellar panel of 4 Kernel Heroes for their time:
- Andrey Lukyanenko | Kaggle: @artgor, Linkedin, Twitter
- Martin Henze | Kaggle: @headsortails, Linkedin, Twitter
- Parul Pandey | Kaggle: @parulpandey, Linkedin, Twitter
- Shivam Bansal | Kaggle: @shivamb, Linkedin, Twitter
Really looking forward to seeing you on the LB and please feel free to leave any questions you might have.