kaggle_age_income_graph.jl

Gadfly.with_theme(:dark) do
	income_col,exp_col = 
	"What is your current yearly compensation (approximate \$USD)?","For how many years have you been writing code and/or programming?"
	income_subset = select(dataset,[income_col,exp_col])

	# Clean out all the missing entries as they are way too hard to deal here
	filter!(exp_col => x->!ismissing(x),income_subset)
	filter!(income_col => x->!ismissing(x),income_subset)
	income_exp_counts = combine(
		groupby(income_subset
			,[income_col,exp_col]),
		nrow=>:num_counts
	)

	# Create the median income columns
	income_exp_counts[!,:median_income] = build_median_income.(income_exp_counts[!,income_col])
	sort!(income_exp_counts,:median_income)
	income_exp_counts[!,:median_age] = build_median_age.(income_exp_counts[!,exp_col])
	sort!(income_exp_counts,:median_age)

	# Get a nice color scheme
	palettef(c) = get(ColorSchemes.algae,c)

	# Plot the graph as a heatmap
	income_by_experience = plot(
		income_exp_counts,
		x=exp_col,
		y=income_col,
		color=:num_counts,
		Geom.rectbin,
		Scale.color_continuous(colormap=palettef)
	)

	# draw(SVG("kaggle_income_by_experience_plot.svg"),income_by_experience)
end