emirhanai / cryptocurrency-prediction-with-artificial-intelligence-v3.0-gru-neural-network

Cryptocurrency Prediction with Artificial Intelligence V3.0 [GRU Neural Network]

Cryptocurrency Prediction with Artificial Intelligence V3.0 [GRU Neural Network]

I am pleased to announce with you the V3.0 of my cryptocurrency prediction project with Artificial Intelligence. I used LSTM in V1.0 and RNN in V2.0. In V3.0, I use the GRU Neural Network, whose algorithms that I have designed myself. The artificial intelligence I created managed to achieve 99.54% accuracy in the March 27 BTC Forecast. Scores are; Total params: 104.658.433, MAE score: 0.02262356696266076, MSE score: 0.0008216340443992631, Accuracy: 98.5220357283052. Another element that I developed in the project is that I wrote algorithm code to visualize neural networks so that people can understand the GRU structure. Everything happens automatically. I am happy to share the project as open source.

Best wishes,

Emirhan BULUT

AI Inventor - Senior Artificial Intelligence Engineer

###The coding language used:

Python 3.9.8

###Libraries Used:

Tensorflow

Keras

NumPy

PyDot

Matplotlib

Scikit-learn (SKLEARN)

Developer Information:

Name-Surname: Emirhan BULUT

Contact (Email) : [email protected]

LinkedIn : https://www.linkedin.com/in/artificialintelligencebulut/

Kaggle: https://www.kaggle.com/emirhanai

Official Website: https://www.emirhanbulut.com.tr

GRU Neural Network Schema [of This Project]

cryptocurrency-prediction-with-artificial-intelligence-v3.0-gru-neural-network's Issues

It appears that the context does not define scaling

How can I solve the problem with scaling and ravel()

I can't find where scaling is instantiated and the ravel() method doesn't work :((

DONT BOTHER IMPLEMENTING THIS

For anyone looking at this and thinking about using it, let me save you some time. This code is broken and will not run like this. The scaler is broken and implemented in two different ways (and will break it running anyway). The neural network also needs modification. Don't waste an hour like I just did trying to fix it, and even when I did it didn't train well.

ValueError: Input 0 of layer "gru_1" is incompatible with the layer

seems like gru layers need "return_sequences=True" to return sequences as output.
https://errorsfixing.com/keras-tensorflow-input-to-rnn-layers/