Developing Data Products: Coursera Project

Classification trees are an important, frequently used method in machine learning.

Changing the parameters used by a specific classification algorithms can strongly affect the resulting tree.

This Shiny app - developed for Coursera's course 'Developing Data Products' - demonstrates the impact of changing various parameters on learning classification trees with the rpart package.

To build the classification tree the iris data set is split into a traing set (80% of the data) which is used to learn the tree and a test set (20% of the data) which is used to predict the species. The tree is build with R's rpart package.

The user can vary two of the parameters using a slider:

• Minimum number of observations per split node
• Minimum number of observations per terminal (leaf) node

The user will see a fitted tree and the confusion matrix of the prediction ( like the one shown below)

            Reference
Prediction   setosa versicolor virginica
  setosa         10          0         0
  versicolor      0         10         1
  virginica       0          0         9

 Accuracy 
0.9666667 

The application itself: https://serviceexplorer.shinyapps.io/ClassificationTree/

The R code: https://github.com/aschniertshauer/IrisClassification