# Context While studying neural networks in machine learning, I found an ingenious 2-D scatter pattern at the [cn231](http://cs231n.github.io/neural-networks-case-study/) course by Andrej Karpathy. Decision boundaries for the three classes of points cannot be straight lines. He uses it to compare the behavior of a linear classifier and a neural network classifier. Often, we are content with the percentage of accuracy of our prediction or classification algorithm but a visualization tool helps our intuition about the trends and behavior of the classifier. It was definitely useful to catch *something strange in the last example* of this description. How does your preferred method or algorithm do with this input?  # Content - The 2D scatter train pattern is a collection of 300 points with (X,Y) coordinates where `-1