June 30, 2025
Day 25 - MORE OF MODEL DEVELOPMENT
What I Learned
Today I continued to make prediction model and increasing its accuracy. Initially on my model, there were some data leakage that was showing 100% accuracy. Later after going back and forth on my code and working on it. I found the cause and was able to fix it. Later I worked on XGBoost and Random Forest. Using XGBoost, I was able to predict how long will the flight be delayed by. After working on XGBoost, I started working on Random Forest and currently working on it. After its completion today, I will still work on evaluating using RMSE, MAE, accuracy.
Blockers
I have had no blockers today.
Reflection
Today’s work was a valuable learning experience in model building and evaluation. Initially, my prediction model showed perfect accuracy due to data leakage, which highlighted the importance of carefully validating the data pipeline. After identifying and resolving the issue through thorough code review, I observed more realistic performance metrics. I then focused on implementing and tuning XGBoost to predict flight delay durations effectively. Following this, I began experimenting with the Random Forest algorithm, which is still in progress. Once completed, I plan to evaluate the models using metrics such as RMSE, MAE, and accuracy to compare their performance. Overall, today was productive and deepened my understanding of predictive modeling and evaluation techniques.