Notebook

In [1]:

using CSV, DataFramesMeta, Statistics

# --- Plotting Functions - Makie ----
# using CairoMakie;
# set_theme!(theme_ggplot2())

# --- Plotting Functions - GadFly ----
using Gadfly
Gadfly.push_theme(:dark)
set_default_plot_size(15cm, 15cm)

Tidy Tuesday Example in Julia¶

Today we'll demonstrate a quick exploration and model that you might perform with TidyTuesday data for practice in Julia.

Getting Data¶

Option 1: Download the CSV from the `tidytuesday` repo¶

Users of Julia 1.6+ can use Base.download to get the data from the subfolders in the official tidytuesday repo and then CSV.read to read it into memory as a DataFrame.

In [2]:

filepath = "https://github.com/rfordatascience/tidytuesday/blob/master/data/2022/2022-03-29/sports.csv?raw=true"
# Declare `missingstring` to convert 'NA' and 'NAN' values to the `missing` type
df = CSV.read(download(filepath), DataFrame; missingstring=["NA", "NAN"])
df |> describe

Out[2]:

28×7 DataFrame

3 rows omitted

Row	variable	mean	min	median	max	nmissing	eltype
	Symbol	Union…	Any	Union…	Any	Int64	Type
1	year	2017.68	2015	2018.0	2019	0	Int64
2	unitid	1.8436e5	100654	181738.0	800001	0	Int64
3	institution_name		ASA College		Yuba College	0	String
4	city_txt		Aberdeen		Yuma	45	Union{Missing, String31}
5	state_cd		AK		WY	45	Union{Missing, String3}
6	zip_text	4.20049e7	604	49104.0	997757500	45	Union{Missing, Int64}
7	classification_code	7.77933	1	6.0	20	0	Int64
8	classification_name		CCCAA		USCAA	0	String
9	classification_other		ACCA		uscaa	130642	Union{Missing, String}
10	ef_male_count	2126.25	0	986.0	35954	0	Int64
11	ef_female_count	2496.21	0	1248.0	30325	0	Int64
12	ef_total_count	4622.46	0	2259.0	66279	0	Int64
13	sector_cd	2.2103	1	2.0	99	0	Int64
⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮
17	partic_women	20.7083	1	16.0	327	63442	Union{Missing, Int64}
18	partic_coed_men	11.0469	1	8.0	130	131560	Union{Missing, Int64}
19	partic_coed_women	14.1669	1	10.0	91	131560	Union{Missing, Int64}
20	sum_partic_men	14.4924	0	0.0	331	0	Int64
21	sum_partic_women	10.8622	0	6.0	327	0	Int64
22	rev_men	809011.0	65	158126.0	156147208	70462	Union{Missing, Int64}
23	rev_women	2.79346e5	0	138318.0	21440365	63444	Union{Missing, Int64}
24	total_rev_menwomen	7.95231e5	130	2.28776e5	156147208	45193	Union{Missing, Int64}
25	exp_men	6.62386e5	65	159666.0	69718059	70462	Union{Missing, Int64}
26	exp_women	3.31594e5	65	141800.0	9485162	63442	Union{Missing, Int64}
27	total_exp_menwomen	7.32422e5	130	234559.0	69718059	45191	Union{Missing, Int64}
28	sports		All Track Combined		Wrestling	0	String31

In [3]:

first(df, 5)

Out[3]:

5×28 DataFrame

Row	year	unitid	institution_name	city_txt	state_cd	zip_text	classification_code	classification_name	classification_other	ef_male_count	ef_female_count	ef_total_count	sector_cd	sector_name	sportscode	partic_men	partic_women	partic_coed_men	partic_coed_women	sum_partic_men	sum_partic_women	rev_men	rev_women	total_rev_menwomen	exp_men	exp_women	total_exp_menwomen	sports
	Int64	Int64	String	String31?	String3?	Int64?	Int64	String	String?	Int64	Int64	Int64	Int64	String?	Int64	Int64?	Int64?	Int64?	Int64?	Int64	Int64	Int64?	Int64?	Int64?	Int64?	Int64?	Int64?	String31
1	2015	100654	Alabama A & M University	Normal	AL	35762	2	NCAA Division I-FCS	missing	1923	2300	4223	1	Public, 4-year or above	1	31	missing	missing	missing	31	0	345592	missing	345592	397818	missing	397818	Baseball
2	2015	100654	Alabama A & M University	Normal	AL	35762	2	NCAA Division I-FCS	missing	1923	2300	4223	1	Public, 4-year or above	2	19	16	missing	missing	19	16	1211095	748833	1959928	817868	742460	1560328	Basketball
3	2015	100654	Alabama A & M University	Normal	AL	35762	2	NCAA Division I-FCS	missing	1923	2300	4223	1	Public, 4-year or above	3	61	46	missing	missing	61	46	183333	315574	498907	246949	251184	498133	All Track Combined
4	2015	100654	Alabama A & M University	Normal	AL	35762	2	NCAA Division I-FCS	missing	1923	2300	4223	1	Public, 4-year or above	7	99	missing	missing	missing	99	0	2808949	missing	2808949	3059353	missing	3059353	Football
5	2015	100654	Alabama A & M University	Normal	AL	35762	2	NCAA Division I-FCS	missing	1923	2300	4223	1	Public, 4-year or above	8	9	missing	missing	missing	9	0	78270	missing	78270	83913	missing	83913	Golf

Option 2: Use `RCall` and the `tidytuesdayR` library¶

We can use the tidytuesdayR R library and your local R installation with the RCall library to call R from Julia.

Example:

YEAR=2022;
WEEK=13;

# TidytuesdayR
tt_data = R"tt_data <- tidytuesdayR::tt_load($YEAR, week=$WEEK)";

# R --> Julia
df= rcopy(tt_data["sports"])

EDA¶

For plotting there's lots of options. I initially used GadFly but switched to Makie recently.

Here's an example with GadFly using the data from 2022 Week 13.

This dataset comes from the 'Equity in Athletics Data Analysis', from Data is Plural So we'll want to make comparisons of sports, colleges, and genders.

Let's look at the kinds of sports there are:

In [4]:

@chain df begin
    plot(x=:sports, Geom.histogram,
        Guide.title("Types of Sports"))
end

Out[4]:

What about gender differences in different sports, by total participation?

In [5]:

@chain df begin
    @rsubset(:sum_partic_women > 0, :sum_partic_men > 0)
    @by(:sports, 
        :partic_difference=(sum(:sum_partic_women) - sum(:sum_partic_men)))
    @orderby(-:partic_difference)
    plot(x=:partic_difference, y=:sports, 
         Geom.bar(orientation=:horizontal), 
         Guide.title("Gender Participation Differences by Sport\n(Left: Skews Male, Right: Skews Female)"))
end

Out[5]:

What about revenues?

In [6]:

@chain df begin
    @rsubset(:sum_partic_women > 0, :sum_partic_men > 0)
    @by([:sports],
        :difference=(sum(skipmissing(:rev_women)) - sum(skipmissing(:rev_men))),
        :abs_difference=abs(sum(skipmissing(:rev_women)) - sum(skipmissing(:rev_men))))
    @orderby(:abs_difference)
    @rsubset(abs(:abs_difference)>1e6)
    last(5)
    plot(x=:difference, y=:sports,
        Geom.bar(orientation=:horizontal),
        Guide.title("Which sports have the highest absolute revenue imbalance?"))
end

Out[6]:

Tidy Tuesday Example in Julia¶

Getting Data¶

Option 1: Download the CSV from the tidytuesday repo¶

Option 2: Use RCall and the tidytuesdayR library¶

EDA¶

Option 1: Download the CSV from the `tidytuesday` repo¶

Option 2: Use `RCall` and the `tidytuesdayR` library¶