About the project
This post will introduce you to a new, open source ai algorithm as well as explain how you can use it. Contra ai is a lightweight, production ready alternative to Tensor Flow. This is written fully in Python and built to solve time series prediction problems, server or client side.
What is Contra AI?
Contra ai is a collections of functions that allow you to solve complex time series problems with AI. If you have used or built or used prediction algorithms before, you will fully understand what is going on under the hood here. Contra is based on the simple KNN algorithm, but with one major modification. Instead of averaging the closest data points, the prediction is based on a linear combination of the sum of all other previous outcomes, and their distance to the requested prediction in terms of their step function.
Why does Contra AI exist?
Other AI engines or libraries can sometimes take minutes to run, and produce weights files that are far to large to run directly on a production website. It is frustrating having to perform complex calculations off-server and then to sending the results over to a database via an API. It is equally frustrating having weights files that are over 0.5 GB large. These challenges make libraries like Tensor Flow useless in solving complex problems fast. Using Tensor Flow for time series prediction is often overcomplicating the problem.
How can You Use Contra AI?
You can use this anywhere you need to solve a complex problem with AI. Contra can be applied anywhere analysis of time series data using AI prediction makes sense. Contra comes with the MIT license, meaning you can use it for personal and commercial projects without worrying about license fees, giving credit, or litigation.
To use this AI to predict stock prices, try it in our web dashboard here. If you are a developer, build with this algorithm in your own apps free with the Styvio API. To access the source code of this project, visit the GitHub repo.