Just found this on the internet and I feel like the #databs community here would find it funny

Finally got my system backup script done/tested for my RPi server.

Next steps, testing/modifying the restore script, adding basic cron jobs for monthly backups/backup cleanups, writing the backups to a microSD, and then will
#selfhost Bitwarden.

#opensource #hardware #homelab #raspberrypi

This is so cool. I am a nerd and now I completely understand why people found programming and hardware and all of it so interesting

Actually so excited for this to finish, just finished creating a Ubuntu Server image, almost the fun part!

Merry Christmas to myself, got a Raspberry Pi 4 and going to be teaching myself some basic hardware stuff!

Will be using Linux, if anyone has suggestions/tips/feedback for learning, please let me know!

#databs #hardware #datacenter

Non-exhaustive example, but should give you a good idea of how I broke things out. Also really helpful when investing because you can then build out different return scenarios to give you an idea of long term $ growth.

Hope this helps, happy to answer any other questions!

Not a bad place to work on a uv tutorial

Then there was Random Forest... 😂

Have a flight tomorrow, so hopefully will get the chance to apply some of what I've l learned and/or do another WhiteWind write up with some recent updates.

Ideally, would love to get a semi-stable model so I can at least submit something by mid-December. #DataBS

Ooof... looks like Linear Regression is not the way to go. Time to try some other things

#DataBS

Wanted to try it out and got this :(

#DataBS Update:

I found a bug with .fetchnumpy, so submitted that. Worked around it using np.array.ravel wrapped around duckdb.sql.df.

LinReg performance was meh, so I tried a KNN classification and was pleasantly surprised. Using 5 neighbors, I was able to produce a ~93.46% test accuracy rate.

Initial model built and tested. Pleasantly surprised by the r2 value here. I'll need to dive in more to interpreting the output.

This is looking at offensive formations and play outcomes. 0.7 RMSE with target values in a 0-7 range, so not horrible for the initial run!

Open to feedback! #dataBS

Making more progress today. Just finished (with the power of ChatGPT) my first "silver" level table. It aggregates the raw plays table to enable better analysis.

Next step, transform/convert these values into numerical equivalents for an initial ML analysis.

#dataBS #DuckDB #MachineLearning

A look at the persistent storage size for DuckDB compared to the flat files for the #NFL Big Data Bowl.

It took a bit to properly format the folder structure and YAML files, but I now have the "bronze" database. Dbt should really shine as I aggregate or split tables for model development. #dataBS

I decided to enter the NFL Big Data Bowl to keep learning and improving my ML portfolio. It’s also a fun way to continue with DuckDB and other ETL tools for data processing!

Currently using dbt-core and DuckDB to load the initial flat files. I think I’ll expand the database later on.