Up to this point we’ve been looking at only two features at a time. We’ve done this largely so that we can visualize the decision boundary. With only two features, the decision boundary is a line in the plane defined by the two features.
The models we’ve looked at so far (Perceptron, Adaline, and Logistic Regression are applicable to any number of features.
Using the Iris dataset, focus on the species Iris-virginica and Iris-versicolor. These two classes are not linearly separable when you use only the two features petal length and sepal length.
Train the Adaline learning model using the following
All six cases of using two features at a time.
All four cases of using three features at a time.
The one case of using all features at once.
Do not use Scikit learn for this assignment. You may, if you want, use the sample code that I’ve posted to Blackboard.
Summarize your results (i.e, what’ s the best accuracy you can obtain for each of the 11 cases you considered) in a table.
Discuss your findings. Does using more dimensions help when trying to classify the data in this dataset?
