Forum Replies Created
-
Posts
-
No, we don’t have census tract boundaries available, just the boundaries listed in the screenshot of the Variables menu in the formula editor. Plus, you would probably want a lot more than just the boundaries, right? I don’t remember where I got the CA data long ago. Sorry about that.
Hi Travis,Here is one just for California. I’ve never tried to make one for the whole U.S. It would be very large, of course.Bill
Hello Jannis,Thanks very much for bringing this to our attention. This is indeed a new problem. I’ll contact Tim Erickson, the developer of the plugins served from codap.xyz.Bill
Hello,Yes, it’s been awhile! But I reconstructed the CODAP document here so you can play with it.The critical move is, once you have the three attributes in place, to go to the graph’s Configuration menu (looks like a histogram) and choose¬†Bar for Each Point.Hope this helps,Bill
As a postscript, I note that CODAP’s map component¬†does have a heat map feature under its ruler menu. See the enclosed screen capture.Would something like that for the graph satisfy your needs?
Hello,Yes, there certainly situations like the one you describe in which it is not clear how many data points are crowded into a small portion of the graph. And CODAP does not have built-in functionality to automatically remedy things.However, I can think of three techniques to get around this limitation. They are explained in this CODAP document and in the enclosed screenshot.Thank you for the suggestion. I’ll log it as a user feature request.Bill
And here is a link to a CODAP document that contains the Fatal Encounters plugin (in case it wasn’t clear how to get to it).
Hello Itai,The CODAP file couldn’t be uploaded with your post. Could you create a shared link for it instead?It might be that the solution to the problem is simply to¬†drag the table rows into the desired order once you have specified that the rows are to be treated categorically.Bill
Hello Fathom fan,
It’s great to hear that you’re using CODAP in your statistics classes, and too bad CODAP hasn‚Äôt been terribly good for classical inferential statistics.But that has been remedied! Go to Tim Erickson’s codap.xyz where, near the bottom, you will find his Testimate plugin. Add this plugin to your document and you’ll find it handles all the inferential stats computations you need for teaching intro stats. (We’ll soon make this a built in plugin for CODAP so it’s easier to find and use.)Let us know of your experience!
Bill
Hello,There is no intrinsic limit, but, in practice 5000 data points is a reasonable limit for use in classrooms. And with many classroom computers even that many points slows things down enough that the students’ experience is adversely affected. When you import a CSV CODAP checks the size and, if it is more than 5000 cases, suggests that you sample rather than import the whole file.On the good news front, the new version of CODAP that we’re working on now will be able to handle¬†much larger datasets.Bill
Hi Tom,We’re hard at work on a new version of CODAP. (Sorry, but no release date available yet.) I’ll log your request, likely in the form that the date format should respect the browser language/locale.Bill
Hello Andrew,Thanks for this excellent excerpt and suggestion. We originally planned to provide this, following the example of TinkerPlots but, somehow, we never got around to it. I’ll write a feature request and link to your post!Bill
Hello, I can answer some of your questions even without access to a shared CODAP document.
1. Nope. No TinkerPlots mixup button. The case plot (the one with randomly scattered points) has a rescale button at the top of the inspector palette, but all it does is re-randomize the point positions.
3. You can stack histograms by dropping a categorical attribute on the¬†right axis.4. a. Unfortunately, you aren’t missing something.
4. b. Not possible currently as you know. Not likely to be a feature we add. Does the case card representation help you at all?
4. c. Have you tried “shift-click” and Cmd-click?
5. I think we chose to report r-squared because of its intuitive interpretation as the proportion of variability accounted for by the linear fit.
I notice that you attempted to upload a CODAP document, but apparently the forum won’t accept that. Try a shared-link instead.
Bill
Hello,I’m sorry to say that CODAP does not have the ability to create the plot you want. If it had a function, let’s call it the “normalQuantile” function, we could do the following:
- Compute the mean, standard deviation and count of the given data at the “parent” level of the dataset.
- Sort the values in the dataset in ascending order.
- Compute the percentile of each value using (caseIndex – 0.5) / count
- Use the normalQuantile function to compute the value we would expect from a normally distributed random variable for each percentile.
- Then we could plot the dataset values on the y-axis and the computed “normal quantile” values on the x-axis and we would have the desired plot.
But, as mentioned above, CODAP does not have the necessary normalQuantile function. But there’s no reason it couldn’t, so I’ll add that function as a desired feature.Thanks for your interest in CODAP!Bill