I had a fun time last Saturday (28 November 2020) joining the National Data Science Challenge 2020 by RISTEK-BRIN and Shopee with Imam Bhaskara, Gaody Mohammad, and Ibnu Yudistira in our Nightrider team. Even though we applied for the Beginners category, we are forced to change to the Advanced category as the committee founded that we’re not suitable in that category (I think it’s because of me, sorry guys). Fearlessly, we joined the competition even though we’re not sure on what will we face because ((( why not, eh? ))) it’s a great opportunity for us to learn.
As expected, we’re surprised when the challenge (that were officially spoiled a week before with dummy data but we were too busy to have a look) was about image processing. We’re challenged in the time period of 3 hours to compare whether two images are the same product, given the dataset of thousands of images and training data of image name 1 and 2, product name 1 and 2, and the label. None of us have the experience to do that, and we’ve just realized it a day before so no time to learn properly. We literally didn’t do anything for the first 30 minutes, but we wanted to save our face by at least submitting an answer and not the last in the leaderboard. As our goal is low, we decided to only compares the product name to see whether two product is the same using Levenshtein Distance (inspired by ertugodabasi code in GitHub). After a few tweaks and adjusments, we finally manage to got a 0.41169 score out of 1 (it’s like 41,1% out of 100%). Phew, we did it! We were happy that we place 93rd (only for a moment because we submitted early), and we were happier to know that we reached our goal of “submitting an answer and not the last in the leaderboard” as we are placed in 135th out of 139 teams submissions (hopelessly assumed that some of the teams didn’t manage to submit so that we assumed that we’re not really that bottom 5). Great job, team!
What I Learned
There are people beyond (this) person, and skies beyond (this) sky.
As the Chinese proverb suggests, “No matter how good you think you are, there is always someone out there that is better.” Although I might thought that my knowledge about data suffices (which I’m not), there are still at least 134 teams of people that is better than me, and I’m sure there are more greater people out there too. When I think that my data visualization and processing is enough, there are still a lot of things to learn about the data field with image processing is one of them. It’s not a reason to be overwhelmed, just enjoy the process.
When there is a will, there’s a way
Even though we thought that we’re doomed (which we actually are), we didn’t give up and still tried to find another workaround to at least achieve the what we called “minimum viable product” that we could deliver. Indeed it’s not the best as we didn’t manage to get to the top 100, but at least we didn’t leave empty-handed. It’s part of the process, and again, just enjoy the process.
Competition is a fun thing
It’s my second experience of joining data competition (the first one was the one held by Telkomsel that I’ve shared in my LinkedIn) and I found that programming in a competition is different with the work programming. Despite of the similarity of time-constrained task, competitions are usually held in a very short time period (a matter of hours) compares to work tasks that needed at least a day to finish. We are expected to gives all of our best effort in that short time period and it pumped out our adrenaline. One thing for sure, I currently don’t want to set a high goal by winning it when joining a competition as I think that winning is a bonus, while the most important thing is to have fun, and again, just enjoy the process.
Congratulations to all winners! There are still a lot of work for me to do to achieve that level, and it will be a great privillege to be honored with it. But don’t forget, it’s not all about winning, but rather an opportunity to evaluate our skills, keep learning, having fun, and just enjoy the process :)