Continuing from my previous post, in this post I will discuss on the inferential and predictive analysis.

About the dataset and the problem to solve: a brief

The dataset is derived from UCI Machine learning repository and the task is to predict if a donor has donated blood in March 2007 (1 stand for donating blood; 0 stands for not donating blood). There are 776 instances in 6 six variables and it is a classification problem.

A. Correlation

As a first measure, I check for strongly correlated predictors. The correlation between two variables is a number that indicates how closely their relationship follows a straight line. correlation refers to Pearson’s correlation coefficient. A correlation of 1, indicates perefct linear correlation. I notice that the predictor total number of donations and total blood donated in c.c are linearly correlated. There is a fairly strong negative linear association between number of donations and months since last donation (corr= -0.159). Next, to visualize the pairwise correlational matrix, I use the pairs.panel() from the library(psych)  which is shown in Fig 1

 library(psych)
pairs.panels(train.data[c("Months.since.Last.Donation","Number.of.Donations","Total.Volume.Donated..c.c..","Months.since.First.Donation","Made.Donation.in.March.2007")]) 

Fig 1: Correlational matrix

 

 

Filed under: classification techniques, inferential statistics, R Tagged: R