July 8, 2025
Day 31 - Working on Model
What I Learned
Today, I worked on increasing the AUC value of our model. I did some research on how to increase AUC value of our model. I did try hyper parameter tuning yesterday but unfortunately my computer couldn’t handle it. So, I did some research. I saw that we can try to make a hybrid model using two or more models. So that is what I did. I searched on how to combine and make them hybrid. Initially, I had some issues becasue I kept having an error. But I fixed it later and worked on making it. With the previous AI model od XGBoost and Random Forest, I made some more models using neural network, lightgbm and catboost. Then I tried making hybrid using Voting Classifier, Stacking Classifier and custom Ensemble. Everything is running right now and I am just waiting for the final output to see which hybrid model has the highest accuracy.
Blockers
It was hard understanding the concept of developing a custom Ensemble model but with some more research and tons of youtube videos, I was able to work it out.
Reflection
It was a great day. I learned about custom Ensemble model development and also was able to work on it. Working on making a hybrid model was a bit tough for me but I was able to get hang of it. I am hoping it get over 90% accuracy with the hybrid model. After completing it, I will be working on exporting charts and all the necessary graphs to use it on our paper.