ML Wiki

Summarizing Data

Before we do any Data Analysis, need to see if data is good

Why?

• Data too big to look at
• Need to find problems before analyzing

Problems:

• Missing values
• Values outside of expected ranges
• Values that seem to be in the wrong units
• Mislabeled variables/columns
• Variables that are the wrong class

Summarizing Data in R

Summary Statistics

• summary(x) - summarizes all quantitative and qualitative variables
• quantile(x) - range of variables

sapply(x[1, ], class)

• calls class for every element of the 1st row
• tells if data was loaded properly

names(x)

• columns' names

Sizes:

• dim(x) - size of the dataset
• same as nrow(x) and ncol(x)
• length(x) and unique(x)

tables

• table(x) - unique + counter
• table(x, y) - two-dimensional table

logical tests

• any(x > 10) - are there any TRUEs?
• all(x > 10) - are all trues?
• which(x > 10) - which elements are TRUEs?
• which(is.na(x)) - which are NAs
• use ! not, & and, | or:
  • which(!is.na(x) & x > 10)
• sum(is.na(x)) - how many NAs

summarizing by columns or rows

• rowSums, rowMeans
• colSums, colMeans

Source

• Data Analysis (coursera)