Inspiration
Our inspiration was notion scheduling and we realized that the notion planner app calendar sucks and we wanted to make a calendar suited for software engineers and other product developers.
What it does
This is a machine-learning program that will determine whether a meeting should be hosted in-person or written in an e-mail instead. We use a decision tree algorithm for a number of reasons: 1.) it's easier to visualize and add/remove from a decision tree model, 2.) decision nodes would be easier to provide metrics about the meeting (like the agenda, purpose, affiliates, etc.), 3.) creating a decision tree closely fits into how a software engineer would determine whether or not the meeting had been summarized on an e-mail.
How we built it
We wanted to improve upon many of the modern-day scheduling and calendar softwares using artificial intelligence (how cool would that be!). This program could be implemented to schedules that deal more with the software development industry. A user can create a "post" on a calendar, and they can add "tags" which pertains to the decision nodes on our decision tree model (see below). After each meeting, an e-mail would be sent to every participant attached with a survey asking "Could this meeting have been summarized in an e-mail? (Yes/No)". Then, the survey data would be accumulated into a .csv alongside the rest of our category data from the scheduling system. The survey data is our "target" which is used for training data.
Our machine learning algorithm recognizes patterns of categories that are grouped together and outputs a value that predicts whether the meeting should be held in-person or be written on an e-mail.
Challenges we ran into
We ran into a challenge with visualizing the product and we ran into some problems with recursion tracing. We also ran into problems trying to create a more efficient webpage since none of us know any javascript and trying to learn it overnight wouldn't be effective.
Accomplishments that we're proud of
We are actually college freshmen, with barely any prior experience in coding before starting college! The fact that we managed to grind out a project like this in a short timeframe shows how much growth and learning we've gone through in this intensive 48-hour time period.
What we learned
We learned we have a long way to go and a lot more to learn especially many mainstream languages such as javascript.
What's next for Comcast Meeting Decision Tree
We would like to build a productivity management software that consists of a calendar system, which would pair nicely with our decision tree algorithm!
Log in or sign up for Devpost to join the conversation.