British Columbia

A line plot is a much better choice here as the time intervals between the data points seems important.

It's suspicious that an entire dimension is not depicted visually.
The reader is forced to calculate the time spans and the implied rate of changes in the two intervals.

The chart doesn't actually use the space on the page to show anything more than text would; it is a waste of space.
They could easily have presented the data in a line graph to show more and make the reader do less.

Sort of valid, in a "when were these households connected, using only 3 data points" kind of way. There's no graph law, so everything is legal and valid. But this is like printing a book on ticker tape, not impossible, just really dumb.

It's not valid. Peach is labeled 77% but occupies 20% of the pie.

This chart with these labels is definitely invalid. The dataset it's based on is coherent, but the labeled pie chart is invalid

This is an instance where a 3d pie graph would be better

As a data analyst / engineer, I usually go through a similar journey producing those percentages.

Boss: "I need to know when we'll reach 100% of households having high speed internet."

Analyst: "Good morning to you too! Who has the data?"

Boss: "I thought you do?"

Analyst: "No, I'm pretty sure I haven't seen any, uh, internet speed data in our warehouse."

Boss: "Have you looked in the computer?"

Analyst: "... I'll talk to Marc, he's got his contacts with the ISPs, we'll figure something out. What does high speed mean?"

Boss: blank stare

Analyst: "Right, I'll talk to the Definitions Department. How should I calculate the forecast? Average over the..."

Boss: "Right, I gotta head to a meeting. You've got until tomorrow EoB."

Analyst: Hectically mashes a dozen different data shapes of historical reports into a unified model, plots the progress so far, curses, tosses some generic trend formula at it because fuck trying to actually figure out the semantics

I've always said bar or line, especially with time axis. In cases like this, you usually use a line as it usually isn't linear growth anyway and you'd use a logarithm to predict where it'd be in 2027.

Of course they're just bullshitting that 100% as real predictions wouldn't ever get there.

You could easily improve it significantly by labeling the segments:

• up to 2017
• 2017-2024
• 2023-2027

I think the way to go would be to give each slice a different radius, so you could see that they're layered over one another.

I (somehow) figured it out. Picture it as 3 separate pie charts stacked on top of each other. Or 57% as the base amount and each subsequent area is the increase on the previous amount.

2017 = 57% 2024 = 20% 2027 = 23%