Exploring grouped data
Data Manipulation in Julia
Katerina Zahradova
Instructor
Filtering for groups
# Companies in different countries
choc_comp_france = filter(chocolates -> chocolates.company_location == "France", chocolates)
choc_comp_brazil = filter(chocolates -> chocolates.company_location == "Brazil", chocolates)
choc_comp_peru = filter(chocolates -> chocolates.company_location == "Peru", chocolates)
choc_comp_belgium = filter(chocolates -> chocolates.company_location == "Belgium", chocolates)
...
groupby() to the rescue!
GroupedDataFrames
groupby() in action
# Group by company_location
groupby(chocolates, :company_location)
GroupedDataFrame with 60 groups based on key: company_location
First Group (156 rows): company_location = "France"
Row company bean_origin ...
String String ...
_______________________________
1 A. Morin Agua Grande ...
...
Last Group(4 rows): company_location = "Ireland"
Row company bean_origin ...
String String ...
_______________________________
1 Wilkie's Organic Amazonas ...
groupby() on multiple columns
groupby(chocolates, [:company_location, :cocoa])
GroupedDataFrame with 373 groups based on keys: company_location, cocoa
First Group (5 rows): company_location = "France", cocoa = 63.0
Row company bean_origin ...
String String ...
_______________________________
1 A. Morin Agua Grande ...
...
Last Group (1 row): company_location = "Ireland", cocoa = 89.0
Row company bean_origin ...
String String ...
_______________________________
1 Wilkie's Organic Amazonas ...
Order matters
Number of records
# Group by country
chocolates_by_country = groupby(chocolates, :company_location)
# Number of records per group
combine(chocolates_by_country, nrow => :count)
60x2 DataFrame
Row company_location nrow
String31 Int64
______________________________
1 France 156
2 U.S.A. 764
3 Fiji 4
...
Sort by number of records
# Group by country
chocolates_by_country = groupby(chocolates, :company_location)
# Sort the number of records
sort(combine(chocolates_by_country, nrow => :count), :count, rev = true)
60x2 DataFrame
Row company_location nrow
String31 Int64
______________________________
1 U.S.A 764
2 France 156
3 Canada 125
...
unique() on a vector
# Unique elements
unique(chocolates.company_location)
60-element Vector{String31}:
"France"
"U.S.A."
"Fiji"
"Ecuador"
"Mexico"
"Switzerland"
"Netherlands"
...
unique() on a DataFrame
# Create a new DF containing only unique rows
unique(chocolates)
1795×10 DataFrame
Row company bean_origin ...
String String ...
_________________________________
1 A. Morin Agua Grande ...
...
unique() with specified columns
# Unique company records
unique(chocolates, :company)
416×10 DataFrame
Row company bean_origin ...
String String ...
_________________________________
1 A. Morin Agua Grande ...
...
# Unique company AND cocoa contents
unique(chocolates, [:company, :cocoa])
968×10 DataFrame
Row company bean_origin ...
String String ...
_________________________________
1 A. Morin Agua Grande ...
...
Let's practice!
Data Manipulation in Julia
