Inspiration
We spend more than a healthy amount of time on Reddit. Of course Reddit has the most valuable currency - karma - which serves to provide reputation to more prominent "Redditors". We realized that their are bound to be some trends that set high-karma posts away from low-karma posts. That is why we built our machine-learning based tool to analyze these trends.
What it does
It looks towards two mediums. One is the number of words in the posts, since shorter or longer titles could have an advantage that must be looked at. The other is the time of posting, cause on a daily basis there are bound to be times where more redditors are online.
How I built it
We had to build 3 components. One was the parser - a tool which browses the web for specific data - we used this to find title length and the time of posting. Another was the actual machine-learning tool, this was based off of the data we parsed earlier and gives a decently accurate prediction of how much karma a given post with some title and time of posting will earn. And then a front-end website that users could use.
Challenges I ran into
We were stuck without an idea for a longtime, thus we fell behind a little. Other than that the biggest error is that a single day is hardly enough time to build an effective machine learning tool. In an ideal scenario we would have more time to access more data and get better results.
Accomplishments that I'm proud of
We went from having no idea how to to make a machine-learning tool, to now feeling somewhat comfortable with the idea. I'd say that's pretty good.
What I learned
We learned new languages a whole new idea to expand what we can do.
What's next for Karmaxer
Spend more time analyzing data to get better results. We would also like to see the web-parser become more efficient because if it is slowed, it may not return any data. Also we would love to see Karmaxer move from analyzing trends from just a day to over the span of a week.
Log in or sign up for Devpost to join the conversation.