I’m trying to create an initial model for my regression but I don’t want the coefficient for Item5 showing up, just the rest of them and Item5_1, Item5_2, etc since it’s already encoded. I also have to follow the K-1 rule and create a dummy variable for the number of categories in the categorical variables I pick. How can I adjust the code to not give Item5 and instead Item5_1, Item5_2, etc?

categorical_variables <- c(
  "City", "State", "County", "Zip", "Area", "Timezone", "Job", "Marital",
  "Gender", "ReAdmis", "Soft_drink", "Initial_admin", "HighBlood", "Stroke",
  "Complication_risk", "Overweight", "Arthritis", "Diabetes", "Hyperlipidemia",
  "BackPain", "Anxiety", "Allergic_rhinitis", "Reflux_esophagitis", "Asthma",
  "Services", "Item1", "Item2", "Item3", "Item4", "Item5", "Item6", "Item7", "Item8"
)

continuous_variables <- c(
  "Population", "Children", "Age", "Income", "VitD_levels", "Doc_visits",
  "Full_meals_eaten", "VitD_supp", "Initial_days", "TotalCharge", "Additional_charges"
)

I then ran anova test and only included variables that are statistically significant, p-value <0.05

Continuous variables

continuous_variables <- c("Children", "VitD_levels", "VitD_supp", "Income") # Included "Income"

Categorical Variables

categorical_variables <- c("Item5" , "HighBlood")

Combine the lists of variables

variables_to_keep <- c(continuous_variables, categorical_variables)

Subset the dataframe to only include these variables

medical_data <- medical_data[, variables_to_keep]

Using these variables as dummy variables because they have 2 or more categories for their data

Create dummy variables for the categorical variables K-1 rule

Would just include 1 dummy variable since ‘HighBlood’ has 2 categories. HighbloodYes would be in the model

medical_data_with_dummies <- dummy_cols(medical_data, select_columns = c("Item5", "Reflux_esophagitis"))

Replacing a variable with over 3 unique categories with one with a relatively low p-value

Extract the names of the dummy variables created, excluding the original variables

categorical_dummy_variables <- setdiff(names(medical_data_with_dummies), names(medical_data))

Now that categorical data has been converted to dummy variables, I can output cleaned data for submission

Made file path for the CSV file

file_path <- "/Users/name/Desktop/medical_clean_forReview.csv"

Filter out dummy variables that only have 1 factor level
Remove columns with only one unique value (dummies with 1 level), to ensure easy analysis

medical_data_with_dummies <- medical_data_with_dummies[, sapply(medical_data_with_dummies, function(x) length(unique(x)) != 1)]

Assuming ‘medical_data_with_dummies’ is your dataframe and it’s already in your R environment

write.csv(medical_data_with_dummies, file = file_path, row.names = FALSE)

Create an initial multiple linear regression model

model <- lm(Income ~ ., data = medical_data_with_dummies)

I tried to remove Item5 from

categorical_variables <- c("Item5" , "HighBlood")

but when I ran

medical_data_with_dummies <- dummy_cols(medical_data, select_columns = c("Item5", "Reflux_esophagitis"))

It says:

 Error in dummy_cols(medical_data, select_columns = c("Item5", "Reflux_esophagitis")) : 
  select_columns is/are not in data. Please check data and spelling.