The spam filter in the lecture had 58 input variables to determine the output variable. Most of these record either the proportion of certain

words in the email ("money" = 0.01 would mean that "money" made up 1% of the words) or the proportion of certain characters in the email

(such as "!" = 0.01 meaning that exclamation points made up 1% of the characters in the email). The variable TOTCAPS, the total number of capital letters in the email, is quite different. When a PCA (principal components analysis) is done with the variables NOT being normalized, the first principal component is dominated by TOTCAPS and captures 92.7% of the variance. When the variables are normalized, the first component captures only 11.6% of the variation and is not dominated by any variable.Explain why this kind of result is to be expected.