MachineHack successfully concluded its fifth instalment of the weekend hackathon series this Monday. The Financial Risk Prediction hackathon was greatly welcomed by the data science enthusiasts with active participation from over 250 participants and close to 500 registrations.
Out of the 265 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: Ashijith Rajendran
Ashijith is working as a Data Scientist at Technosoft Global Services. He did his Post Graduate Program in Business Analytics from Praxis Business School. His keen interest in mathematics helped him develop a passion for data science, a passion which he chose to make his career. He regularly updates his skills by reading articles and participating in hackathons.
Approach To Solving The Problem
Ashijith explains his approach as follows:
I started by visualising the data. Then, I created some of the generic models using Logistic Regression, Random Forest, XGBoost and Support Vector Machine. I chose Random Forest Classifier, hoping that it could give good results when fine-tuned. I started tuning the parameters and performed cross-validation, but the results showed no improvement. I went on to perform data exploration to identify the hidden patterns in the data, which helped me in generating new features. I used the generated features on the Random Forest model that got me the best score and the winning solution.
“MachineHack is a good platform for every data scientist. It gives an opportunity to implement the learning and compete with other fellow data science enthusiasts,” he shared his experience.
Get the complete code here.
#2: Kranthi Kiran
Kranthi Kiran is a Computer Science engineering student at Army Institute of Technology, Pune. He first came in touch with machine learning when one of his friends was doing the Titanic Survival Challenge and was amazed that he could predict the survival of a person in a natural disaster.
This made him curious enough to try out the problem on his own. By the time he finished the competition, he was totally astonished by the power of analytics and machine learning on real-world problems.
Approach To Solving The Problem
He explains his approach briefly as follows.
On EDA, I found that there were some rules for which the records could be directly classified as 1 (target). I found 5 rules which almost covered 48% of the data given to predict.
- Internal_Audit_Score >= 9 ==> Target = 1
- External_Audit_Score >= 9 ==> Target = 1
- Final_Score >= 9 ==> Target = 1
- Loss_score >= 9 ==> Target = 1
- Past_Results >= 2 ==> Target = 1
Next, I made some features based on numerative operators like the addition of scores, subtraction, multiplication and chose 3 best features that boosted my Local CV a little bit.
I baselined almost all models of which gradient boosting methods worked great, especially CatBoost. I ended up using CatBoost for my final model, which worked a tad bit better on the local CV than any other model.
“MachineHack is a great platform for anybody practising data science and machine learning as you can compete with anybody starting from a student to a data scientist with 10 years of experience and learn tremendously in parallel to competing with the best,” he shared his opinion on MachineHack
Get the complete code here.
#3: Niranjan K
Niranjan is a 2017 Mechanical Engineering Graduate. He started his career with a core MNC, which provides mechanical design solutions to its clients. Soon, he found out his job to be redundant. Therefore, he started exploring new domains and got acquainted with data science and machine learning. As he kept exploring more and more about this emerging field, he found out to be in line with his interests. Having been inclined towards automation, he always wanted to be in a position where he could directly influence any business. Thus, he quit his job to pursue a postgraduate program in data science from Great Lakes Institute of Management. From then on, he has been honing his programming and data science skills.
“Being from a non-programming background, I never doubted my ability to learn to code quickly. The key is to never stop learning,” he said.
Approach To Solving The Problem
Niranjan explains his approach briefly as follows:
- Correlation plots suggested there was very less multicollinearity among independent features.
- The KDE plot of each feature gave insight into the distribution of data.
- Variable importance using Gradient boost suggested that two features (Internal_audit_score & Fin_score) had 76% weightage in predicting the risk factor.
- I evaluated logloss scores for different models using H2O AutoML and Gradient Boost gave the minimum logloss.
- Tuned the hyperparameters manually and got the best score on the leaderboard.
“This is my third hackathon in MachineHack, and I got into the top 3 in two of them.The experience has been amazing, and I look forward to many more such learning experiences through MachineHack,” he shared his MachineHack experience.
Get the complete code here.
