small updates to cleaning and modeling

code/data_cleaning.R

@@ -111,7 +111,7 @@ nrow(data)
 
 # Create choice data ---------
 
-# First gather the data
+# First convert the data to long format
 choiceData <- data %>% 
     pivot_longer(
         cols = cbc1:cbc8,
@@ -136,5 +136,18 @@ choiceData <- choiceData %>%
 
 head(choiceData)
 
+# Create new values for respID & obsID
+nRespondents <- nrow(data)
+nAlts <- max(survey$altID)
+nQuestions <- max(survey$qID)
+choiceData$respID <- rep(seq(nRespondents), each = nAlts*nQuestions)
+choiceData$obsID <- rep(seq(nRespondents*nQuestions), each = nAlts)
+
+# Reorder columns - it's nice to have the "ID" variables first
+choiceData <- choiceData %>% 
+    select(ends_with("ID"), "choice", everything())
+
+head(choiceData)
+
 # Save cleaned data for modeling
 write_csv(choiceData, here("data", "choiceData.csv"))

code/install.R

@@ -0,0 +1,11 @@
+# Install packages - only need to do this once!
+
+# Install packages from CRAN
+install.packages(c(
+    "fastDummies",
+    "here",
+    "janitor",
+    "logitr",
+    "remotes",
+    "tidyverse"
+))

code/modeling.R

@@ -23,10 +23,10 @@ data <- clean_names(data)
 
 # Estimate the model
 model <- logitr(
-    data   = data,
-    choice = "choice",
-    obsID  = "obs_id",
-    pars   = c(
+    data    = data,
+    outcome = "choice",
+    obsID   = "obs_id",
+    pars = c(
         "price", 
         "type_fuji", "type_gala", "type_honeycrisp", "type_pink_lady",
         "freshness_average", "freshness_excellent")