Machine Learning Image Forecast

We will give you: 1. Satellite image every 5 minutes across the days. 2. Actual value every 15 minutes across the days. Just bid on it, and then I will send you the paths. You will build a machine learning tool to map historical satellite cloud images to values, which will build the training model. You will then use the model you trained to predict the value for the next three hours. Please be aware that there is also a seasonal pattern plus the cloud pattern. For example, summer has a higher actual value than winter. You don’t have to forecast off-peak values. Benchmark: you need to predict for the next hour with at least below 10% error to get paid; otherwise, I am better off extrapolating the line and still beating your model. I will pay particular attention to 08:00 to 14:00. • Bidding Your bid is my budget. If you deliberately put in a low bid to attract me to speak to you, you either revise your bid, or I will ignore you. • Delivery We need the source code and will integrate your program into a much larger one. You will be paid once you finish all the work and it works on my machine. We will also attack your program, as it has to be secure and reliable without loopholes. It has to be bug-free. • Platform Your code has to work on Windows. • Workplace You will be working online remotely. • Experience Please show us your detailed related work experience and portfolio. Without similar experience, you are unlikely to be awarded, as you may not finish the project anyway. • IP The intellectual property, including the source code, belongs to Sors Markets Ltd. • Interview We may interview you to see if you can work on the project. • Demo We will need to do a demo on Freelancer. If you have any questions, please send us a message. • Further work More work may need to be awarded to the winner.