Prediction model system with UI, Django, Posgresql.

I'm wanting a prediction program made. With a really nice UI and lots of cool features. It would be used for a variety of uses and would generate it's data from a prediction poll using a pool of users. This data could be entered manually, or imported in, or later be fed to database with a scraper. It would allow for a group of individual bots/agents that make predictions on the outcome of events. I'm wanting to record and analyze the predictions from these agents to see if they know more than they think they know, or know more in combination somehow with other agents to make more accurate predictions together than they could on their own. It could potentially use weighted data or give more points to a certain agents prediction on an event because of past history in a similar situation. For example: Agent 1 predicts that it will rain today, Agent 2 predicts that it won't rain today. However from our historical data we can see that Agent 1 isn't very good at guessing when it is going to ran when guessing on Saturdays for whatever reason and only has a 25% accuracy out of 50 tracked predictions. And Agent 2 predicts that it will rain with a greater percentage accuracy and so far out of 100 predictions has guessed that it would rain with 75% accuracy. Agents 3-50 also have their own prediction records for if it will rain today or not. I want to be able to track the accuracy of each agent for each outcome over time and build models to play around with. It would track things like how often each agent predicted correctly for each day of the week, and how often they picked a particular outcome. For example Agent 3 might always say it's going to rain on Wednesdays but is only right 55% of the time. There could be hidden data that might show that Agents 2 & 7 predicts overall with accuracy of 75% each. However on days that they both match up and say the same thing then it rains 85% of the time. django/python 3.6 with bootstrap built on rest api, Nginx, react.js, responsive site, green unicorn. (not all tech maybe required for this job) Depending on which version you use you may need install it in a virtual environment using virtual EnV all of our server are running Ubuntu/Linux Server 18.04 or 16.04 at the earliest No Docker. You may use our server for development or not. Either way it needs to be installed as part of this job when completed. Relational database if required use Posgresql only. Please do not edit main server files, just change settings in your virtual environment only.