And coders don't really like surprises! (Unless a pick you generate happens to match the draw results...)

Tested all 14 games against the universal classifier script in 3 modes

1. With offset and window set to 0

2. With offset of 0 and window of 21

3. With offset of 10 and window of 21

This was to ensure that all games work, AND that the data slicing was correct when introducing non zero values for the offset and the classification window.

This allows the GUI development to finally begin!

Copies of all files, including the csv history files, have been placed in a project folder, git enabled (real version control, not just my ad hock file naming convention!) And basic documentation started.

Next up is to build the "hello world" in Kivy to ensure the libraries are properly installed... then it is off to the races!

Will start with the classifier and updater scripts... adding followers when ready. Still trying to add a temporal component to that, idea well is running a little dry... not exactly the same as "most recent appearance" in the classifier, but similar in nature.

The placeholder splash screen will just have text to start, followed by the main screen. Sub screens will have dynamically generated grid layouts based on each game.

Once the windows app is perfected, then it is on to a Docker container and Linux to package the app as an APK for Android. Because of the myriad of problems that can arise in a mobile app, it is imperative that the windows version operates flawlessly, including sanitizing input values on the settings page to avoid "index out of range" errors... I am gonna try my best to break it!

This will be the project until done, no specific time line... it is more important that it works! Then the goal will be met, a way to get picks without needing to ever touch the laptop... though there are still a whole host of projects waiting there, from moving from follower output to a first order Markov Chain to studying vertical sums with both code AND Excel powerPivot in the pick 3 to putting data into a real database where SQL commands can be used to find patterns and trends that I can't seem to spot... that WILL require the laptop...

2026 is looking pretty well organized as far as hobby coding goes, much to learn, much to do.