Monday Also Quiet

I had no plans for Monday, and my sleep was deep and ended near 8. I might have woken a few times, but the 1/2 Benedryl had stopped the itchiness I sometimes get and put me out, and I have no memory of rising. I don’t like to use things like Benedryl as they often have a hangover that makes it hard to rise, but it still helps when my allergies expand to hives or just being uncomfortable. The tired I felt Sunday was gone, as were the hives. Better!

A year back, rising for work at the shoe company was described more like Dracula floating out of his coffin to discover it was not night. Ugh! Now, I wake with joy that I have another day. With Susie’s passing, my brain tumor recovery, being retired by Nike, and cancer with chemotherapy, I am always happy and often amazed to have another day and to awake well and ready to find what will be included in this bonus day. While unbidden, even depression and sadness are welcome, as it still means I have another day. Good morning!

Joy! I make joy in my French press with coffee that reminds me that we can help others by buying coffee that does not take advantage of poor farmers in South America, Equal Exchange band. The coffee in my cup is pure, dark, and bitter, reminding us there is much more to do. On the red bag of coffee, it says I am a “citizen consumer,” and my choices matter. I feel hope and caffeine when I taste liberal coffee.

With the kitchen found, it has not moved, and coffee made, I start on the next blog. I also text a few good mornings while I am writing, read emails, read the news (political and about the war in the Middle East), and update my Quicken with the latest transactions. Quicken downloads most transactions (saving bonds, 401K, and CD interest are done manually, usually every month) and helps me ensure no unplanned (theft) activity on the accounts, and I (and the tax authorities) know how my money flows.

I fry bacon with enough extra to make BLT later. I crack two eggs, each time in a small bowl, and pour the egg in the hot bacon grease. The non-stick pan is perfect, and the eggs are soon loose in the hot fat. I swirl them around carefully and splash a little bacon goodness over them. I carefully flip the mostly cooked eggs in the hot grease for an over-easy finish. A banana and two eggs over-easy with bacon are breakfast with my cup of pure liberal joy.

I spend the early morning writing the blog, reading, and texting. After the blog is done, I dillydally and surf the Internet, but soon decide I need to dress. Dressing is how I mentally accept that I have things to do during the day; otherwise, I could stay in PJs (or less) all day and do nothing. Dressed, I started writing code on the Kaggle website, remembering my Python and Pandas.

Aside: Pandas is an extensive add-on library for Python. It is a database-like add-on that lets me easily manipulate data to align it for acceptance by AI handling routines from yet another library: sklearn (scikit-learn is the organization name). All of this uses the super fast and older numpy, which reminds me of the FORTRAN math libraries I used years ago in college. Pandas, sklearn, and numpy are the ‘trinity’ of AI coding and the brilliant work of thousands of developers (often, these libraries are updated together to improve their interactions and speed), and all are Open Source. Yes, nobody owns them.

I managed a tiny bit of progress in my code. I have plenty of time on the coding contest and have much to relearn and discover. I review an example and am sure the code is poor and wrong. I take a long walk in the neighborhood, gain 3,000 steps (4,600+ for the day), and rethink the code while walking. I used to do this at work and even back at college; walk and code in my mind. The steps create a beat that pushes your mind. I am sure the example code is questionable and broken. I am also concerned that I am including an error in my design from the example, a shortcut I have made before that is cheap but could fail. I will have to accept that risk for now. I will look for a means to mitigate or remove it.

I return to The Volvo Cave and rest for a while, reading and napping. I rise again and open a 16oz can (1/2 size) of Bush’s Orginal Baked Beans with bacon and brown sugar (it is an American thing), plus a piece of toast to eat with it (not British style with the beans on the toast) for lunch. It was good. I bought a case of these cans because I liked them; they are inexpensive on Amazon, and this covers me for a week’s supply of food for a disaster that requires nothing more than a can opener (I have a manual one, a battery-powered one, and an electric can opener).

Back to coding and learning how to use Pandas and sklearn routines to solve issues in the data set I have been given in the contest. While not full of mistakes and text issues, it takes me all afternoon to devise a cleaning of the data. I finally loaded the data into an AI model, and it failed. Time to make dinner!

I was talking to Deborah on my iPhone when I remembered I had an eggplant left and saw it was fading fast. Soon, I imagine the expiring eggplant would rise in a cheap black-and-white horror movie, become an undead veggie, and slap me around. I rang off and addressed the issue. Soon, the threat was resolved with the pieces of the veggie soaking in saltwater with a plate and a cup pushing them under water. No undead rising today!

I found a stir-fry recipe but had no fresh ginger, so I added onions and garlic. While it was pretty plain in flavor, it was eggplant, after all. Dinner was not terrible, with some jasmine rice made to go with the stirfry adding more flavor. Next time, I needed even more garlic and spices. Also, I would soak the eggplant in milk to add more flavor to the veggie.

I watched more Slow Horses, season 4, on Apple+. The new season has all-new stories (the books ran out in season three), and I like it, but it does feel different. It is more subtle, and I like that.

I’m back to coding, but to do that, Wildwood Taphouse would better suit my mood. Air Volvo gets me there without effort. Soon, I have a red ale, pretzels, and Kaggle on my laptop, sitting on a metal stool. Dave is the bartender, and I have not met him before; I learned he usually works the other Wildwood on the other side of Beaverton. I coded the changes needed to align the data so there were no holes. Data frames allow for a no-data or null condition, but AI models need a value. I have to impune values. I look at examples and adopt their strategies but use, I think, a better organization to the code. I also keep to my process of not changing the data supplied to train the AI model but aligning instead a copy of the original. Mistakes happen and are hard to spot if you have changed your starting point.

Aside: I do not have a pronoun for Ryan in the following story, which means you will find Ryan’s name used, not a pronoun; it can be repetitive. I will not assign a gender when a name is what I have. I take it as a challenge to make this not read stilted.

I use the sklearn data splitting process and train my model. A few mistakes mean more beer and pretzels. While others would see this as work, I am enjoying every moment.

A chef I have met, Ryan, sits beside me at the bar and nearly spits out the supplied beer when I tell Ryan I am coding AI. I, more beer helping both of us, cheerfully explain how and why the AI works. I showed my fellow bar sitter the data and how I aligned it. I expect an 80% solution once I tune the model. It’s not good enough to win the contest, but it’s fair. The chef is elated to have someone demonstrate how and why this works.

Ryan is fascinated by my AI work and its possible application to cooking. We talk animatedly about the parameters for bread making and pizza crusts. Ryan is an expert at pizza and Italian-style food and serves as a chef at a local restaurant. My fellow beer drinker explained that the recipes that were perfected in Las Vegas, Ryan’s previous gig was in the desert city, did not work in Portland as the air, flour, and water are different here, meaning cooks and chefs must change recipes to make a good pizza and bread here from something that works in Las Vegas. Ryan wondered if we had accurate measurements that an AI could adjust a pizza crust recipe for various locations. Is there a decision tree and data that could make this an equation? Ryan had much to think about while enjoying more beer.

Thinking of AI coding and my model at least being accepted in my code, pay the bill, say good night to Ryan and Dave, and head home after boarding Air Volvo. More code to evaluate the model is next. Home without incident, I read and soon showered and dressed for bed. I managed to sleep without any assistance from painkillers or Bendril.

Thanks for reading.