Module 8 Exercise - Fitting Basic Statistical Models: Part 1

Setup

Load Required R Packages

Tidyverse will be used for handling data processing/manipulation, and the here() package assists in pathing. Tidymodels is used for model handling.

Load Data

# Pathing
data_path <- here("fluanalysis", "data", "SympAct_Any_Pos.Rda")

# Data Importation
raw_data <- readRDS(data_path)

Data Manipulation

During this initial stage, unnecessary columns and any NA values will be removed according to the assignment instructions. Column removal will be accomplished through the syntax select(!contains()), for this method includes all columns that are not the ones specified in the command. Tidyverse has the command drop_na(), permitting the removal of all observations containing any NA values across the dataset.

# Removal of Unnecessary Columns
clean_data <- raw_data %>%
  select(!contains(
    c("Activity", "Score", "FluA", "FluB", "DxName", 
      "Total", "Unique.Visit")))

rmarkdown::paged_table(clean_data)

# Remove NAs in Dataset
clean_data <- clean_data %>%
  drop_na()

rmarkdown::paged_table(clean_data)

Save Cleaned Data

As specified in the assignment instructions, the resulting dataset shall be saved as an RDS file in the data folder of fluanalysis. This allows for loading of the cleaned data in another Quarto/R script.

# Pathing
data_path <- here("fluanalysis", "data", "cleaned_data.rds")

# Saving Dataset as RDS
saveRDS(clean_data, file = data_path)

Addendum: Exercise for Week of 3/27/2023

Feature Removal

Weakness, Cough, and Myalgia exist on both a severity score and as Yes/No. They may confuse the model, so all Yes/No versions will be removed.

clean_data_March <- clean_data %>%
  select(!CoughYN) %>%
  select(!CoughYN2) %>%
  select(!WeaknessYN) %>%
  select(!MyalgiaYN)

Categorical/Ordinal Predictors

For an easier time building and handling the model, categorical variables will be coded as unordered factors and others as ordered factors.

Categorical Variables

# Recipe Creation
categorical_recipe <- recipe(~ SwollenLymphNodes + ChestCongestion + ChillsSweats + NasalCongestion + Sneeze + Fatigue + SubjectiveFever + Headache + RunnyNose + AbPain + ChestPain + Diarrhea + EyePn + Insomnia + ItchyEye + Nausea + EarPn + Hearing + Pharyngitis + Breathless + ToothPn + Vision + Vomit + Wheeze + Weakness + CoughIntensity + Myalgia + BodyTemp, data = clean_data)

# Categorical Variables
categorical_dummies <- categorical_recipe %>% 
  step_dummy(SwollenLymphNodes) %>% 
  step_dummy(ChestCongestion) %>%
  step_dummy(ChillsSweats) %>%
  step_dummy(NasalCongestion) %>%
  step_dummy(Sneeze) %>%
  step_dummy(Fatigue) %>%
  step_dummy(SubjectiveFever) %>%
  step_dummy(Headache) %>%
  step_dummy(RunnyNose) %>%
  step_dummy(AbPain) %>%
  step_dummy(ChestPain) %>%
  step_dummy(Diarrhea) %>%
  step_dummy(EyePn) %>%
  step_dummy(Insomnia) %>%
  step_dummy(ItchyEye) %>%
  step_dummy(Nausea) %>%
  step_dummy(EarPn) %>%
  step_dummy(Hearing) %>%
  step_dummy(Pharyngitis) %>%
  step_dummy(Breathless) %>%
  step_dummy(ToothPn) %>%
  step_dummy(Vision) %>%
  step_dummy(Vomit) %>%
  step_dummy(Wheeze) %>%
  prep(training = clean_data)

# New Dataset Generated
categorical_dummies <- bake(categorical_dummies, new_data = NULL)

Ordered Factors

# Ordered Factors Using Previous Objects Generated
ord_recipe <- recipe(~ SwollenLymphNodes_Yes + ChestCongestion_Yes + ChillsSweats_Yes + NasalCongestion_Yes + Sneeze_Yes + Fatigue_Yes + SubjectiveFever_Yes + Headache_Yes + RunnyNose_Yes + AbPain_Yes + ChestPain_Yes + Diarrhea_Yes + EyePn_Yes + Insomnia_Yes + ItchyEye_Yes + Nausea_Yes + EarPn_Yes + Hearing_Yes + Pharyngitis_Yes + Breathless_Yes + ToothPn_Yes + Vision_Yes + Vomit_Yes + Wheeze_Yes + Weakness + CoughIntensity + Myalgia + BodyTemp, data = categorical_dummies)

# Change Expected Ordinal Variables to Ordered Factors
categorical_dummies$Weakness <- as.ordered(categorical_dummies$Weakness)
categorical_dummies$Myalgia <- as.ordered(categorical_dummies$Myalgia)
categorical_dummies$CoughIntensity <- as.ordered(categorical_dummies$CoughIntensity)

# Ordinal Variables
ord_dummies <- ord_recipe %>%
  step_ordinalscore(CoughIntensity) %>%
  step_ordinalscore(Myalgia) %>%
  step_ordinalscore(Weakness) %>%
  prep(training = categorical_dummies)

# Generate New Dataset
ord_dummies <- bake(ord_dummies, new_data = NULL)

Removal of Near-Zero Variance Predictors

filtered_process <- ord_recipe %>%
  step_nzv(all_predictors()) %>%
  prep(training = ord_dummies)

# Final Cleaned Dataset
clean_data_March <- bake(filtered_process, new_data = NULL)

Saving Copy of New Data

This newer cleaned dataset shall be saved as well to be used in the “machinelearning.qmd” document.

# Pathing
data_path <- here("fluanalysis", "data", "cleaned_data_March.rds")

# Saving Dataset as RDS
saveRDS(clean_data_March, file = data_path)