Thanksgiving Dinner
Survey data, Network graphing
Notable topics: Survey data, Network graphing
Recorded on: 2018-11-20
Timestamps by: Alex Cookson
Screencast
Timestamps
Exploratory bar chart of age distribution (and gender) of survey respondents
Using count function on multiple columns to get detailed counts
Parsing numbers from text using parse_number function, then using those numbers to re-level an ordinal factor (income bands)
Exploring relationship between income and using homemade (vs. canned) cranberry sauce
Adding group = 1 argument to the aes function to properly display a line chart
Rotating text for axis labels that overlap
Getting confidence intervals for proportions using Jeffreys interval (using beta distribution with an uniformative prior)
Explanation of Clopper-Pearson approach as alternative to Jeffreys interval
Using geom_ribbon function add shaded region to line chart that shows confidence intervals
Using starts_with function to select fields with names that start with a certain string (e.g., using "pie" selects "pie1" and "pie2")
Using gather function to get wide-format data to tidy (tall) format
Using str_remove and regex to remove digits from field values (e.g., "dessert1" and "dessert2" get turned into "dessert")
"What are people eating?" Graphing pies, sides, and desserts
Using fct_reorder function to reorder foods based on how popular they are
Using n_distinct function count the number of unique respondents
Using facet_wrap function to facet food types into their own graphs
Using parse_number function to convert age ranges as character string into a numeric field
Exploring relationship between US region and food types
Using group_by, then mutate, then count to calculate a complicated summary
Exploring relationship between praying at Thanksgiving (yes/no) and food types
Empirical Bayes binomial estimation for calculating binomial confidence intervals (see Dave's book on Empirical Bayes)
Asking, "What sides/desserts/pies are eaten together?"
Calculating pairwise correlation of food types
Network graph of pairwise correlation
Adding text labels to nodes using geom_node_text function
Getting rid of unnecessary graph elements (e.g., axes, gridlines) with theme_void function
Explanation of network graph relationships
Adding dimension to network graph (node colour) to represent the type of food
Fixing overlapping text labels using the geom_node_text function's repel argument
Tweaking display of percentage legend to be in more readable format (e.g., "40%" instead of "0.4")
Summary of screencast