To explain it further, you can think about PCA as an axis-system transformation. PCA is a technique that takes a set of correlated variables and linearly transforms those variables into a set of uncorrelated factors. In plain English, what is principal component analysis (PCA) in Excel? In practice, we often encounter correlated data series: commodity prices in different locations, future prices for different contracts, stock prices, interest rates, etc. Are there hidden forces (drivers or other factors) that move those 5 variables?.If we were to use those variables to predict another variable, do we need the 5 variables?.The five (5) variables are highly correlated, so one may wonder: To better understand the problem, let’s compute the correlation matrix for the 5 variables: Note that the scales (i.e., magnitude) of the variables vary significantly, so any analysis of raw data will be biased toward the variables with a larger scale, and downplay the effect of ones with a lower scale. First, we place the values of each variable in a separate column and each observation (i.e., census tract in LA) in a separate row. Data Preparationįirst, let’s organize our input data. Each observation represents one of twelve census tracts in the Los Angeles Standard Metropolitan Statistical Area. The five variables represent the total population (“Population”), median school years (“School”), total employment (“Employment”), miscellaneous professional services (“Services”), and median house value (“House Value”). ![]() In this tutorial, we will use the socioeconomic data provided by Harman (1976). Next, we will closely examine the different output elements in an attempt to develop a solid understanding of PCA, which will pave the way to more advanced treatment in future issues. In this tutorial, we will start with the general definition, motivation, and applications of a PCA, and then use NumXL to carry on such analysis. This is the first entry in what will become an ongoing series on principal component analysis (PCA) in Excel.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |