Improve plotting with seaborn to scale better

[niklas-heer]

Currently, I'm using seaborn with Python to display the results. But if a lot more languages are added, the details being displayed will become too much. I'm looking for help with displaying the results in a better way, preferably with Python and seaborn. Maybe switching to Matplotlib would be now appropriate?

speed-comparison/analyze.py

Lines 26 to 86 in df56f35

	def plot(df, rounds, to_file):
	# Theme
	sns.set(style="dark", context="paper")
	sns.set_color_codes("pastel")
	plt.style.use("dark_background")
	# Plot
	# TODO: Find a way to display the platte scale to highly accuracy better.
	bar = sns.barplot(
	x="average",
	y="name",
	data=df,
	width=1,
	edgecolor="black",
	linewidth=2,
	errwidth=0,
	log=True,
	# TODO: Improve color palette.
	# https://seaborn.pydata.org/tutorial/color_palettes.html
	palette=colors_from_values(df["accuracy"], "light:seagreen"),
	)
	bar.bar_label(
	bar.containers[0],
	fontsize=10,
	padding=3,
	)
	plt.xlabel("Average time (ms) in log scale", fontweight="bold")
	plt.ylabel("Language", fontweight="bold")
	# Title
	plt.suptitle(
	"Speed comparison of various programming languages\n",
	fontweight="bold",
	fontsize=20,
	y=1.02,
	)
	plt.title(
	f"Method: calculating π through the Leibniz formula {rounds} times",
	style="italic",
	fontsize=16,
	y=1.02,
	)
	# Caption
	url = f"https://github.com/niklas-heer/speed-comparison"
	plt.figtext(
	0.5, -0.1, url, wrap=True, horizontalalignment="left", fontsize=12
	)
	timestamp = f"Generated: {datetime.now().strftime('%Y-%m-%d %H:%M')}"
	plt.figtext(
	0.2,
	-0.1,
	timestamp,
	wrap=True,
	horizontalalignment="right",
	fontsize=12,
	)
	sns.despine()
	plt.autoscale()
	plt.savefig(to_file, pad_inches=0.2, bbox_inches="tight", dpi=200)

[joelandman]

I've not used seaborn. Will look at that. I noted that the first digit on the top set of results seems to be truncated.

I'll play with this later tonight after work, and see if I can help with it.

[niklas-heer]

Thank you, @joelandman that is much appreciated. Seaborn is as far as I've seen just a Matplotlib which understands dataframes.