There isn’t much of an update to be honest, but I'll explain a little bit more later. Again the over arching theme this project has presented to myself is how rusty I am with python. But I finished the “simple” version of the program with somewhat surprising results, at least in my mind. I took the bracket for the 2019 lacrosse tournament, ran it through the program and it's accuracy was between 46-53% accurate. What this teaches me is that going straight off of which team has the best stats will only get you so far when trying to figure out who is going to win.
So the question is what is the biggest factor in predicting the winning time? That's what I've been thinking about during the day on my vacation from my 9-5 and I'm sure it's what everyone thinks about that tackles this same problem. And I haven't made any head way I'm sorry to say.
Regardless of my progress in solving the above question I’ve started the next phase of the project. I’ve decided on a way to simplify the data I’ve collected and then send it on to train the model. I don't know if it's exactly the right way but I'm still a newbie so I don't know any better and I'm gonna try it anyway. In the data I have 14 stats for each team and game, I decided to follow the example of Rodrigo Nader as he tried to predict World Cup winners. (the article is https://towardsdatascience.com/using-machine-learning-to-simulate-world-cup-matches-959e24d0731 and it's really an interesting read). So anyway I'm subtracting the stats from each team, if the result is positive it means team 1 had the better stat, else team 2 had the better stat. Pretty much it sounds like this will be inline with how I'm doing the "simple" prediction but eventually will be more machine learning-ish.
I’ve been frustrated with this project. I’ve been discouraged and wondered if I should even keep going on with it. In those times I simply take a break. That's the thing I've learned over the past couple months with these side projects, if my hearts not in it, take a step back, take a deep breath and put in on the back burner for a little bit. That's why there hasn't been a ton of progress with it. I hit a wall with python and trying to figure out other things with it has been making me pull out my hair. So I put the project off for a day or two and calm down a bit.
So that's the progress this week on the project. If you want to see where I'm at you can check out the project on my GitHub. And if there's any thoughts, comments, or constructive criticism feel free to let me know!
Top comments (0)