Okay, so yesterday I was messing around with some data related to the danzica airport – you know, Gdansk Airport in Poland. Thought I’d share what I did, ’cause maybe someone else finds it useful, or at least mildly interesting.
First off, I grabbed some public data. I think it was from some open data portal, something about flight schedules and airport operations. The data was a mess, of course. Dates all over the place, weird column names, the usual headache. So, I started by cleaning it up. Used Python with Pandas, ’cause that’s my go-to. Renamed the columns to something I could actually understand, converted dates to a consistent format, and handled missing values. There were a bunch of NaNs in there, mostly for delay times. I just filled them with zeros, figured it was better than dropping the whole row.
Next, I wanted to see some basic stats. Like, how many flights are there each day? Which airlines are the busiest? So I grouped the data by date and airline, and calculated the number of flights. Nothing too fancy, just your standard groupby and count operations. I even threw in a little visualization using Matplotlib. Made a couple of bar charts showing the top airlines by flight volume. Looked pretty neat.
Then, I decided to dig a bit deeper into delays. I calculated the average delay time for each airline and for each route. That was a bit more involved. Had to convert the delay times from strings to numbers, and handle some edge cases where the delay was negative (probably data errors). I found out that some airlines were consistently worse than others when it came to punctuality. Not naming names, but let’s just say I wouldn’t be surprised.
After that, I experimented with some simple machine learning. Tried to predict flight delays based on the airline, route, day of the week, and time of day. Used scikit-learn for that. Started with a simple linear regression model, but the results were terrible. Then I tried a random forest, and that was a bit better, but still not great. I think I need more data and more features to get a decent model. Maybe weather data would help?
Finally, I threw everything into a little dashboard using Streamlit. Just a few interactive charts and tables, so I could easily explore the data and play around with the different features. It’s nothing fancy, but it’s good enough for my own use. Plus, it makes it easier to share my findings with others.
It was a fun little project. I learned a few things about data cleaning, data analysis, and machine learning. And I got a better understanding of what’s going on at danzica airport. Maybe next time I’ll try to build a real-time flight tracking app. That would be cool.
