Capstone Adventures: Part 17

And we're live again!

http://bit.ly/ccrlfost42 - click the Voila button on top to run.

If you're not on mobile and/or the Voila button doesn't work, click the "Not Trusted" button to set to "Trusted". Then with the menu click Kernel -->> Restart & Run All

I'm taking the long weekend off and will begin the paper next week.

Capstone Adventures: Part 16

Programming is done. It launches on Heroku but will only deploy with the $250/month dynos so I think I will keep it a desktop app and switch the csv files to pulling from the parent folder and whatnot. I'll figure out how to rebuild it as something deployable at a later time. Until then, this will not end up in my portfolio.

I have a web dev course on udemy that I can work on after I pass capstone. Hopefully that will shed some light on what I'm doing wrong (other than using a Jupyter notebook). I also have the goat book which I should have worked on during my term break.

Back to desktop I suppose. I'll need some kind of logging mechanism for security purposes.

Capstone Adventures: Part 15

I feel like it's been a while since I had good news to share. I got the more advanced features to run on Voila with the exception of a confusion matrix heatmap that's supposed to generate on command. Oh well. I just switched the output to a text that makes the numbers appear in a grid. Though I'll play with refresh able heatmaps later today.

The markup and formatting are also done or at least mostly (I'm always tweaking). Hopefully I'll get it deployed in the next few days. Heroku is my first stop.

Also, my extension was approved so this thing is not due on Sunday. I have until the end of September but I'm hoping to turn it in within the first week of September.

Capstone Adventures: Part 14

My app is now fully functional. It has everything it needs to be an app except it's still a jupyter notebook. Voila broke a few of the more advanced functions so I'm attempting a few other options today. I think if it's not working by the end of the day I'll remove the advanced functions, deploy it on Heroku, and call it a day ... it's getting harder to count how many ways my capstone keeps failing.

I do have some good news. My capstone extension was approved. So I have until the end of September to pass task 2. However, if I can get it done by the end of the thee-day-weekend next week, I'll take that win and go back to this in the future. I do not want to be working on this beyond the 8th.

Capstone Adventures: Part 13

Just a short update. Qgrid refuses to work on my computer. My husband was kind enough to set up my environment on his computer but there's a slight change that part isn't going to work when it's time to launch on Heroku.

I'll worry about that later. Interactive queries done.

Up next, I need to make my data visualizations interactive. Back to looking stuff up.

Capstone Adventures: Part 12

PyQt5 broke again and doesn't seem to want to cooperate with Spyder so I think it's time to put the desktop GUI to bed for good.

I learned enough Flask and Django to know that I don't know enough to get something going on those. Javascript plays a role and I I don't know javascript. There are also a few other pieces I don't grasp well enough to take what I have and implement it with these tools.

Good news though. It looks like I can use Voila to integrate most of my jupyter notebook with a few modifications. I'd basically create a few widgets to make it functional use Heroku to host the notebook as a single web page.

I still have to learn a few things to get this to launch, but I should have a basic implementation of most features by the weekend.

Capstone Adventures: Part 11

I was told that a desktop gui app is far more complicated than a web app, and also that I probably should have read the goat book and looked into Django (would have been nice to know that a month ago).

This is the kind of thing a mentor should be able to give advice about. I've been teaching mine about the course more than he's been teaching me. He has no clue what he's on about half the time and when I do ask him a question, I do so in the hopes that he asks someone for me because he has access to the graders --even though he rarely thinks to use that resource.

If you're not willing to learn about the course you are tasked to instruct, you probably shouldn't be an instructor for that course. The bar for this one is way too low for our capstone, WGU. My mentor didn't know what data cleaning was. Can looking up all the terms on the rubric be a minimum requirement for being a capstone instructor?

Anyway, back to work. I will work on this until I hear back about my extension. If I don't get it, I'll stop to read a few books, then build the app on a few different platforms so I can report on the various needs and methods for taking on the different approaches. You guys can get the diploma in my place.

Capstone Adventures: Part 10

Time is running low. I have the GUI drafted though not fully functional. I'm chipping away at all the pieces but I keep getting this nagging feeling that I'm making it harder than it needs to be.

The rubric was holding me back a bit. I tried to get clarification on a few things and my CI was clueless. So I had to build the various options to figure out the answer. That really tanked my motivation, especially after I figured out what I needed to know and how easy it would have been for the CI to simply read the capstones that have been submitted so far (what are we at, 20 graduates by now?).

In any case, I applied for an extension and if I don't get it or finish by the end of the term, I will withdraw. While I can afford another term, WGU is starting to not feel worth it. At this point, I would rather give that win (not to mention tuition) to a school that actually helps me get there --even if I have to pay more for that privilege.

Capstone Adventures: Part 9

I made a lot of progress once I switched away from Computer Vision topics. I am now determining Cervical Cancer Risk based on a data set in the UCI Repository. The model will be based on it but the data used can be collected by just about any clinic, hospital, state, etc.

One of the more interesting features I implemented was customizing the risk threshold so that the user can select their acceptable levels of false negatives vs. false positives. It took a while to figure out how to convert the continuous variable back into binary in a format that allows me to compare the results to the original test data. But I finally finished that today.

I'm ready to start integrating all the features into a GUI. I didn't set up security features or queries because it didn't make sense to do that in Jupyter, but I left all the other stuff in there because it doesn't make sense to put all of that into the GUI.

So Jupyter is done. I even printed it out as a pdf to ceremoniously close that segment up. I kept it open on my desktop until today so it's literally closed as well, lol.

I'll spend the next day or two exploring GUI options before digging into that part. Hopefully all that time in Software 1 and 2 transfers over into doing similar things in Python.

I really have no idea how much time I need to finish. I guess I'll find out once it's time to start the next phase. After this, I just have the writing. There are 28 days left in my term.