Machine learning test repo

prophet:

Forecast and prediction based on historic Bitcoin trade data. Data cleanup/consolidation and removal of outliers.

tensorflow:

Softmax classification of the Iris dataset with Tensorflow. Loading the dataset and also encoding the one-hot label manually for better understanding. After the dataset training we generate random test sets and run them to get either probabilities and predictions from the learned data.

Visuals:

sklearn:

Simple Nearest Neighbors algorithm used in combination with the Iris dataset. Even if our dataset is small, the ball tree classification is used. We generate random test sets and get predictions based upon uniform and distance weighting.