Decision boundaries learnt by training with some R library classifiers

  1. A 2-D non-separable dataset (generated uniformly and randomly) is used to train and generate the decision boundaries with a few of the following R library classifiers.
  2. The following are the decision boundaries learnt by different ML models from a few R packages, when trained on the above dataset.db.png
  3. The following are the decision boundaries learnt a few linear classifiers such as (logistic regression, with / with L1/L2 penalties, linear discriminant analysis) and also with Naive Bayes.
    db-simple.png
  4. The following are the decision boundaries learnt by k-nearest neighbor algorithm with different values of k.
    db-knn.png
  5. The following are the decision boundaries learnt by a backpropagation neural net with single hidden layer with different numbers of hidden units.
    db-nnet.png
  6. The following are the decision boundaries learnt by the CART decision tree with different values of the pruning parameter.
    db-rpart.png
  7. The following are the decision boundaries learnt by the random forest ensemble classifier with different numbers of trees.
    db-rf.png
  8. The following are the decision boundaries learnt by the support vector machine classifier with different values of the C parameter.
    db-ksvm.png
  9. As can be seen, the models overfit / underfit the training data with different values of the parameters used for learning.
Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s