Passion is like genius; a miracle.

Install tree package: Let’s check how iris looks: Build tree: Let’s check how it looks: There are 6 terminal nodes in the tree. Let’s plot it: See how accurate it is: We have pretty small misclassification rate. But, to avoid overfitting, let’s do k-fold cross-validation and prune the tree: Deviance (entropy like measure used in…

First, remove species: Draw hierarchical clustering: It’s very tempting to pick three clusters as we already know that there are three species. So, cut cluster at three: Now we got three clusters for the iris data. We can check it’s accuracy using table: It’s easy to see that versicolor is often confused with virginica in…

Don’t use pairs, but splom. This renders: Reference) R: Scatter Plot Matrices. http://stat.ethz.ch/R-manual/R-devel/library/lattice/html/splom.html

There are two useful functions for clustering: hclust (for hierarchical clustering) and plclust (plotting cluster). Given a matrix: Here, “average” is the way we compute distance between clusters. In case of average, we compute the average distance between every possible pairs (a, b) where a and b is from two different clusters. That’s easy part.…

Look. You should call summary on the ret. val of aov to get the following statistics. Because P value is very small, H0 is rejected; Petal.Width is different depending on Species. We can draw a boxplot to visualize this: Some texts[1] explain that F statistics in ANOVA is (between class variance) / (within class variance)…

Continuing from the previous thread, Let’s compute mahalanobis distance using the equation of it instead of using mahalanobis function: We can compute the distance between two means of two groups using pooled covariance matrix in mahalanobis distance as well.

Mahalanobis distance computes distance of two points considering covariance of data points, namely, mahalanobis distance = (d – AVG(d)) / Covariance = d’C-1d where d is euclidean distance between two points. In R[1]: Now we compute mahalanobis distance between the first data and the rest. Click here for the next article on this topic. References)…

Principal component analysis (PCA) is a mathematical procedure that uses an orthogonal transformation to convert a set of observations of possibly correlated variables into a set of values of uncorrelated variables called principal components[1]. PCA tries to reduce dimension while maintaining axis which has the largest variance. And the problem boils down to getting principal…

Factor analysis assume the following[1]: x – mu = LF + epsilon where L is factor loading, and epsilon is error terms(or uniqueness that’s not explained by common latent factor F). This is basically saying that x is generated from low dimensional F. F is multiplied by L (so that it can be high dimension),…