Tweaking_Quakes

I'm ~really thrilled~ pretty happy with [Streamlit](Streamlit.io). I have not seen something this good looking and easy to work with since folium. It's what I hoped Dash would be. As part of getting to know it, I'm redoing - or at least adding to - Quakes. Essientially, a web app and I may go on from there; I have a few ideas of things I'd still like to add.

Part of this involves a few small but important changes. One is swiching from data formats from pickle to parquet. It's not critical to do it all. But pickle is not secure - the documentation says in with a [friendly warning in a red box](https://docs.python.org/3/library/pickle.html) yet that's not an issue in this case. And it's quite a bit faster than a .csv but much slower than a parquet file. Also, a parquet file will let you make a dataframe from just the columns you want to during import rather than requiring you to filter the entire dataset. There just isn't really a win for pickle in this case other than an arguement for the status quo.

I've been thinking about my preferred data stack for a while now. And my thinking is it's time to shift away from pandas. There are a lot of options Pandas 2.0, Polars, Dask, Rapids and I'm sure there are a few more. But I don't want to jump for the sake it. A large part of my decision making calculus are things like interoperabilty, aestheics, a sensiable interface, and performance.