Code to load data from survey on predicting height using other body measurements

predicting-predictions.R

# data loading and tidying
##########################
# Note: we haven't talked about how to do this, and I don't expect you to
# understand the code below.
library(dplyr)  # useful function for cleaning up data
library(lubridate)  # useful for working with dates and durations
library(magrittr)  # useful for data manipulation
library(tidyr)  # to gather the data
library(googlesheets)  # used to load the google sheet data
sheet <- gs_key("11pXvnrDfygcaMp6iI-30BasveKYLZ4Ce9Z0-cuuKLww") # register google sheet
df_raw <- gs_read(sheet)  # load sheet
df <- df_raw %>% 
  gather(measurement, prediction_value, age:calf_circumference) %>%
  select(-other, -comments) %>%
  mutate(measurement = reorder(measurement, prediction_value),
         pols_209 = ifelse(pols_209 == "No", "Not in POLS 209", "In POLS 209")) 
# done with tidying  
###################  

# quick look at data
tibble::glimpse(df)

# compute the average for each measurement
smry <- summarize(group_by(df, measurement), 
                  average_score = mean(prediction_value),
                  percentile_25 = quantile(prediction_value, .25), 
                  percentile_75 = quantile(prediction_value, .75))

# scatterplot
library(ggplot2)
ggplot(df, aes(x = measurement, y = prediction_value, label = name)) + 
  geom_text(alpha = 0.5,
            position = position_jitter(width = 0.1, height = 0.1), 
            size = 3) + 
  facet_wrap(~ pols_209) +
  coord_flip() + 
  theme_bw()

# plot of averages
ggplot(smry, aes(x = measurement, 
                 y = average_score,
                 ymin = percentile_25,
                 ymax = percentile_75)) + 
  geom_point() + 
  geom_linerange() + 
  labs(x = "Measurement",
       y = "Subjective Predictive Value",
       title = "Average Subjective Predictive Value and Interquartile Range")

# density plots
ggplot(df, aes(x = prediction_value, fill = pols_209)) + 
  geom_density(alpha = 0.5) + 
  facet_wrap(~ measurement, scales = "free_y")

# bar plot
ggplot(df, aes(x = prediction_value)) + 
  geom_bar() + 
  facet_wrap(~ measurement)