Chapter 6: Feature Selection and Dimensionality Reduction

Activity 11: Converting the CBWD Feature of the Beijing PM2.5 Dataset into One-Hot Encoded Columns

1. Read the Beijing PM2.5 dataset into the DataFrame PM25:

   PM25 <- read.csv("PRSA_data_2010.1.1-2014.12.31.csv")

2. Create a variable cbwd_one_hot for storing the result of the dummyVars function with ~ cbwd as its first argument:

   library(caret)
   cbwd_one_hot <- dummyVars(" ~ cbwd", data = PM25) 

3. Use the output of the predict() function on cbwd_one_hot and case it as DataFrame:

   cbwd_one_hot <- data.frame(predict(cbwd_one_hot, newdata = PM25))

4. Remove the original cbwd variable from the PM25 DataFrame:

   PM25$cbwd <- NULL

5. Using the cbind() function, add cbwd_one_hot to the PM25 DataFrame:

   PM25 <- cbind(PM25, cbwd_one_hot)

6. Print the top 6 rows of PM25:

   head(PM25)

   The output of the previous command is as follows:

   ##   No year month day hour pm2.5 DEWP TEMP PRES   Iws Is Ir cbwd.cv cbwd.NE
   ## 1  1 2010     1   1    0    NA  -21  -11 1021  1.79  0  0       0       0
   ## 2  2 2010     1   1    1    NA  -21  -12 1020  4.92  0  0       0       0
   ## 3  3 2010     1   1    2    NA  -21  -11 1019  6.71  0  0       0       0
   ## 4  4 2010     1   1    3    NA  -21  -14 1019  9.84  0  0       0       0
   ## 5  5 2010     1   1    4    NA  -20  -12 1018 12.97  0  0       0       0
   ## 6  6 2010     1   1    5    NA  -19  -10 1017 16.10  0  0       0       0
   ##   cbwd.NW cbwd.SE
   ## 1       1       0
   ## 2       1       0
   ## 3       1       0
   ## 4       1       0
   ## 5       1       0
   ## 6       1       0

Observe the variable cbwd in the output of the head(PM25) command: it is now transformed into one-hot encoded columns with the NE, NW, and SE suffixes.