Forum Replies Created
-
Posts
-
Hello Anna,
Somewhat unfortunately, the documentation for importing CSV data through the URL is somewhat buried in CODAP’s github site. You can find it here.
There is another gotcha however, namely that CODAP runs as a “secure” site with. As such, the site accessed by the URL parameter must also be secure. So the URL you would use will need to have the ‘https’ prefix as in
https://codap.concord.org/app/?url=https://mywebsite.com/fjdkfjk.csv
Here’s what Jonathan Sandoe has to say in an email to another user:
The issue is that the link is to an insecure http address and CODAP is running with an https URL. Browser rules (quite reasonably) prevent a secure page from loading an insecure file. I tried modifying the URL to “https://xyz.com/Sample.csv” but the server doesn’t appear to support https.
If you are affiliated with the owners of xyz.com in any way you could request that they support HTTPS. It’s probably a good idea for them anyway.
The other network bugaboo that limits CODAP’s ability to read data from a third party sources are same origin policies (CORS). Servers can be configured to restrict the ability of browser apps to read their data. If a CSV file’s server is configured in this way, CODAP will not be able to use it directly.
No, sorry, there isn’t. (We have in mind a future feature that will make capturing such things easy.)
But you can use the camera menu to open the graph with the draw tool to add the value by hand.
Hi Stephanie,
Yes, there are ways to massage dates from the bewildering variety of formats in which they appear into formats that CODAP can deal with (without having to type them in by hand.
In the case of the Salinity data, it didn’t prove too hard, though there is one apparent CODAP bug that gets in the way. I’ve updated the CODAP document so it now shows both datasets and has some numbered instructions that came from my efforts. For convenience, here they are.
- Rename the two attributes.
- Set the first attribute Type as Date.
- Set the second attribute type as numeric.
- Delete the first case.
- Make the scatterplot.
- Note that (for some reason) the date scale only goes through 2015. With your mouse cursor on top of Oct, click and drag to the left so that the rest of the points come into view.
I’d say you’re getting a taste of what data scientists and other data folks confront every day.
Bill
Hi Stephanie,
As I suspected, the problem was turning the MONTH attribute into a valid date.
Here is a CODAP document with the data properly munged.
I was able to do it with formulas, but it would have been a lot quicker and easier to just type them in. I erased all the formula work because it was so awkward.
Hope this helps,
Bill
Hi Stephanie,
Yes, there is a fix. I just need to see the data to describe it.
Unfortunately I couldn’t access your CODAP file on Google Drive. But you can upload the CSV file here.
Or you can go to the top-left menu and choose Share followed by Get Shared Link. Then post the link here. I’ll be able to view and edit a copy of your document without changing the master.
Thanks, Bill
Hi Stephanie,
I wonder if the time attribute is not being correctly interpreted? Could you either provide a link to a shared view of your CODAP document, or upload a CSV file of your data so we can take a look?
The “Connecting Lines” option is only available when you have a scatterplot; i.e. when both axes have a numeric attribute. I suspect that your time attribute is being treated as categorical so what even though your graph looks “sort of” right, it’s actually a dot plot and not a scatterplot.
Bill
Hi Josh,
It is possible, of course to add quadratic fitting. That starts us down a slippery slope, though, that includes polynomial fitting and exponential/logarithmic fitting. All that functionality can be pretty off-putting to non-math-geeks. So we’d have to figure out how to provide it without cluttering things up. Maybe in a plugin?!
Bill
Hi Steen,
Jonathan is on vacation this week, but next week he’ll set you up for Danish. Thank you ever so much for being willing to give this a try!
Bill
Hi,
The following should work:
2.39*x-5.6
2.39*Day-5.6
Or you could create two sliders, a and b, adjusted to -5.6 and 2.39. Then you would plot
b + a*Day or
b + a*x
I hope that helps,
Bill
I’m sure you’re right about the problems that having two logins can pose. I’ve seen that, too. Thanks for posting, including the solution!
Sorry, the sampler doesn’t support those input options. But I can see why you might want them! I’ll enter a feature request.
Thanks for pointed out that the MAD is part of middle school common core. I’ll enter feature requests for both displaying it on a graph and for a function to compute it.
Definitely a bug! And you get first call on it. Thanks very much. And nice workaround.
Deleting the measures on Clear Data was by design with the idea that you would be moving on to a new model-making activity. But you’re not the first to wish it were different, so perhaps we’ll rethink. A workaround is to use “Delete All Cases” in the table menu as that leaves all the attributes intact.