Open jlstro opened 6 years ago
Speed promise broken again - sad Failing as a provider - An algorithmic approach 10 serial killers that are better people than my internet provider Here are five things I found out while testing my internet speed over a year - number 3 made me cry
These headlines are killing me loll. Also remind me not to piss off anyone who can code.
And I hear your pain. I also had a dataset that's very datetime-heavy (is that even a word/phrase?), it was Project 1 and I had no clue what to do and my final draft looks real bad.
Then @sarahslo gave me some very very good advice. Why not plot your data in heatmaps?
Like this:
You can read her comment for more info. In fact, you most definitely should.
What I did was using a Python library called calmap to build heatmaps, then clean them up in Illustrator.
This is amazing and hilarious. (I feel your pain with Verizon... ISPs are the bane of my existence.)
It seems like there are some pretty significant dips in your line chart -- it may be worth looking into why speeds were so much slower in April and October. It'd be funny to annotate the break in the chart where your vacuum cleaner ruined everything.
The weekday bar chart is confusing; the bars are all very close in length, the days of the week aren't annotated, and I'm not sure why the bars are in a gradient of colors. I'm not sure what the chart is meant to illustrate for the reader.
It might be worth scaling your charts so you can see just how far your speeds are from 100 Mbit/s.
I'm sure this is going to turn out awesome!
I got a bit lost in a bit of a DateTime
disaster, but here's a quick update:
Disregard the legend for this thing please
I will rephrase most of what I wrote in the final story. I realized that this is not a story, actually, but more a personal revenge thing. I can still see how this would be relevant for more people than me - especially if it could be used as an example to collect more data from multiple households in a similar way and compare the results to the promises that providers make.
DateTime
is a pita
How did you make the heat map? Do you know excel? In excel, if you have a dataset and you select it and you are on the home tab and you choose conditional formatting you can copy and paste the cells into illustrator.
you can then delete all the numbers, and you'll have those data made as a heat map...
you can then take the numbers out. (select>fill color. select>same fill color, delete)
for the line chart you don't need to zero-base it. you are comparing it to itself! you probably want x axis lines so we can see the data a little better.
Headline: What up to 100 Mbit/s really means Published website version: https://jlstro.github.io/Data-Studio/code/03-speedtest/page/index.html Code repository: https://github.com/jlstro/Data-Studio/tree/master/code/03-speedtest Final data set(s): see above
Working with DateTime
kept me busy. I wanted to learn how to handle this in the first place, this was one of the ideas when I came up with the project. I know more about it than before, so that worked.
The initial scope was not very good. There is no general application for what I did and not much to learn for the reader - it is more a personal thing for myself. I would change this if I had to do it again and maybe open the story up, find more general questions like "how honest are companies in their advertisements?"
Whooo sorry I'm so late with my comment! I think you're being too hard on yourself about this story. Personal experience, especially consumer experiences, are a great way to discover stories! I agree that one datapoint (your home) isn't much, but this is cause enough to further investigate internet speeds promised v. actual internet speeds. You could take this further by enlisting others to do the same, either with the same or different internet providers. You could see whether there are websites that test speeds for individuals and then share the data (I know there are websites that test data speeds, I don't know if they keep what they find and whether they make that info public). If it were possible to see data speeds by location (the equivalent of a zip code, or even city by city) you could look at which cities/neighborhoods get the best internet, etc. Or, you could use this as a basis for explaining why it's legal for companies to promise "up to X" and then fall way way short.
A few data points you could pull out and highlight: what was the absolute top speed you recorded? Because that could be a good way to frame the story: the internet provider promises up to 100 Mbits but never actually delivers more than up to 60, so that should be what they advertise. (Follow up question: are there different tiers that promise up to 60 that cost less? Because then you really definitely are getting ripped off!!) And, re the color scheme, I think the red/orange/green is one commonly used for speed tests, and might be easier to understand than the blue scheme. I didn't really know what to make of it, and although I assume darker is faster, you don't actually include a key to explain that clearly.
Anyway, great work, I think this could be a great story, even if it's not quite ready to go as-is. Thanks for sharing!
Pitch
This started out as a personal vendetta against Vodafone, my internet provider. I don't want to go into too much detail here but let's just say that for technical reasons, I am forced to buy my internet from them at home. I am also forced to buy a up to 100 Mbit/s connection, which NEVER EVER delivered 100 Mbit/s EVER in my life. So I went back and forth with them and they said they cannot do anything about it and they are legally fulfilling their part of the contract.
At some point I decided to start measuring my internet speed on to collect some data so I set up a raspberry pi and made it test the speed once every hour. And because I was angry, I also set up a twitter bot that sends out a tweet per day to blame them publicly (It has 6 followers, 5 of them forced by me to click the follow button and myself). I thought it was funny back then, but now I'm not sure, maybe it is pathetic. Anyways, I never did anything with the data I collected so here I am now with roughly 14.000 speed test results from the past year and half. There is a bigger gap from earlier this year, because the raspberry pi (it lives under my couch) was hit by the vacuum and detached from the power cable and I did not notice it.
Summary
The data set includes info on upload speed, download speed and the test server. I'll try to analyze whether there are any peaks or lows in my internet speed across days of week, month, etc.
Details
Possible headline(s): Speed promise broken again - sad Failing as a provider - An algorithmic approach 10 serial killers that are better people than my internet provider Here are five things I found out while testing my internet speed over a year - number 3 made me cry Data set(s): File 1 File 2 Code repository: Here Possible problems/fears/questions: Most of the operations are with the
datetime
data I have and so far it gives me a horrible timeWork so far
Made some basic calculations Speed over time, grouped by week (the gap is due to the vacuum incident) average speed for each day of week
Checklist