MachineHack successfully concluded its eleventh installment of the weekend hackathon series last Monday. The Power Plant Energy Output Prediction hackathon provided the contestants with an opportunity to revisit the very basics of data science and machine learning problem. The hackathon was greatly welcomed by data science enthusiasts with over 300 registrations and active participation from close to 200 practitioners.
Out of the 200 competitors, three topped our leaderboard. In this article, we will introduce you to the winners and describe the approach they took to solve the problem.
Out of the 200 competitors, three topped our leaderboard. In this article, we will introduce you to the winners and describe the approach they took to solve the problem.
#1| Devrup Banerjee
Although Devrup learned python just out of the sheer need to automate the routine work and gather data at scale, his real enthusiasm and passion for data science sprouted in his second year of MBA at Great Lakes Institute of Management, Gurgaon, while he was attending his marketing and retail analytics class. He realized that the real motivation behind learning all these algorithms was not about enhancing accuracy but to tell your client by how much you can promise to increase their bottom-line if they were to follow your exact given path. The subject changed his life.
My college GreatLakes Institute Of Management has been instrumental in introducing me to this field. Till my graduation, I didn’t even know such a field existed and one can even take analytics as majors. It was only during my MBA years that I was introduced to this highly intriguing field. When I was young I kept hearing people say follow your passion, only then you can achieve success, and finally, when I was kept up at night due to problems related to datasets, where maybe I wasn’t able to calculate a certain metrics or reach a certain accuracy, I ultimately understood the meaning of following of passion. I have achieved some kind of success in this field albeit minor compared to the ocean that data science is but then am only getting started. I honestly feel our school and undergrad education system needs to change to at least give students the chance to have a look at the full catalog of subjects available which are so much more relevant professionally, for one to ultimately find their real passion in education.
“Machinehack is doing a great job organizing the hackathons and data science summits which not only provides exposure to budding data scientists and students like me but also helps in opening up networking opportunities with the very best in the domain. The practical approach always is better than theory and machine hack is one of the leaders when it comes to that” – Devrup shared his opinion about MachineHack
Approach To Solving The Problem
Devrup explains his approach briefly as follows:
The train and test set contained lots of duplicates. Now some of those duplicates in test exactly matched those of train. So I grouped the values in train by mean of target and manually imputed all the exact matched by this means in the test. This alone gave me a score of 2.30 on the leaderboard. The remaining values in the test set were predicted by a stacking model between CatBoost and LightGBM. The automated FeatureTools was used for feature engineering purpose and after some solid feature selection this gave me a good score.
#2| Akash PB
Akash did his bachelor’s in Statistics and Masters in Operations Research from IITB. From his college days, he was interested in programming algorithms from scratch and hence came to this field. On one fine day, he found a precious gem on Coursera – Andrew Ng sir’s course, and after completion of his course, he felt a little bit confident to apply stuff in the real world. The company he worked for gave him the opportunity to be a part of a lot of projects that involved machine learning and data science and hence he earned a real-life flavor of theoretical knowledge. Since March, he has been participating in MachineHack hackathons and from there I am learning even more.
“Machinehack is a nice place for me to experiment with things. Moreover, the interface is very nice now as compared to the previous one. The hackathons are nice and hopefully, we will see more challenging stuff in the future.”- Akash shared his opinion.
Approach To Solving The Problem
He explains his approach briefly as follows:
In this hackathon, data was small and hence the best solution would be from blending only. I used xgboost and blended 4 models of xgboost with different parameters. The weights of blending were not done manually. I made sure that the model giving the best RMSE (least RMSE) in a fold will get a higher weight and used exponentiation for that. Furthermore, I did some EDA and created a new feature by dividing ambient pressure with Vacuum.
#3| Tamil Selvan
Selvan did freelance digital marketing during his college days without knowing the term itself. Because that was a time where the term digital marketing is not in the hype. In his final year, he got an opportunity to market a restaurant in Chennai.
It was when he came to know that automation is helping the restaurant industry to make a better experience for customers in many aspects. He researched a lot, got love towards data science machine learning, and AI. My consistent desire to learn more brought me to pursue Data Science in Imarticus Learning, Chennai.
Now He is developing a no-code automated machine learning platform, where one can do EDA, preprocessing, modeling, and gives reports without coding.
“Machine hack is a wonderful platform for beginners. I really love to apply my theoretical knowledge on their hackathons conducting every weekend. Happy to be a part of the talented machine hack community”- Selvan shared his opinion.
Approach To Solving The Problem
He explains his approach briefly as follows:
No high skewness in the response variable, so no transformation is required. He created a new feature by dividing the Pressure variable with Vacuum. Another feature he created was a Dew point. It is the temperature to which air must be cooled to become saturated with water vapor.
He used unsupervised learning to create labels, where he used these labels as predictors for my model. Yet, another feature was created which is the difference between Relative humidity and temperature. Finally, he used a 10-fold XG boost with max depth and learning depth as parameters.
