Blog Post 8
Andrew Y -
Hello everybody! I know you have all been on baited breath and waiting on the edge of your seats for the next update in this riveting project. Well, here it is. This week, some serious changes and shifts happened in the data collection. For the first time in weeks, the A.I. prediction algorithm performed better than the human prediction algorithm, not only in the one week trial, but in the two week trial as well. If this trend does continue into the final weeks, then there might be some reason to believe that the A.I. is actually learning as these trials go on. This isn’t necessarily good for the project, because the assumption was that the A.I. prediction model was already trained. However, there is a possibility that the A.I. prediction model is accurately trained, but is simply using the current data set I am giving it, and modifying its predictions accordingly. This also presents another avenue for further research, because it would be interesting to see how the two models perform after a certain training period. So far however, both prediction models are par for the course: performing with horrendous accuracy compared to the actual market, which is totally expected. I’m nearing the end of my data collection process and in a week or two, I’ll be ready to discuss the data analysis and conclusions. So make sure to stay tuned for that and I’ll see you guys next week!
- Andrew