| 1 |
| Annie, 12:39 (14.06): joins the room |
|
| 2 |
| Annie, 12:41 (14.06): leaves the room |
|
| 3 |
| Annie, 12:30 (15.06): joins the room |
|
| 4 |
| Annie, 12:43 (15.06): leaves the room |
|
| 5 |
| gerirose, 14:09 (15.06): joins the room |
|
| 6 |
| | Bill, 14:39 (15.06): joins the room |
|
| 7 |
| | | cynthia, 14:41 (15.06): joins the room |
|
| 8 |
| | | cynthia, 14:44 (15.06): Hello, |
|
| 9 |
| | Bill, 14:41 (15.06): Hi. Not ready to chat yet. Soon. | |
|
| 10 |
| | | | Beth, 14:49 (15.06): joins the room |
|
| 11 |
| gerirose, 14:52 (15.06): I'm just lurking since I didn't do my homework. I just wanted to say hi to Bill. | | | |
|
| 12 |
| | Bill, 14:47 (15.06): Hi Geri! | | |
|
| 13 |
| | | | Beth, 14:56 (15.06): Hello everyone. I will be joining you for this session. hi Geri, Cynthia and Bill |
|
| 14 |
| gerirose, 14:54 (15.06): Hi Beth! Warming up for your trip to Fathom Summer School? | | | |
|
| 15 |
| | Bill, 14:48 (15.06): You've still got 10 minutes! And it doesn't take that long to do! @: Message 11: To whole message | | |
|
| 16 |
| | | | Beth, 14:57 (15.06): I can not wait!!! |
|
| 17 |
| gerirose, 14:55 (15.06): I'll be out there at the end of your week. | | | |
|
| 18 |
| | Bill, 14:49 (15.06): Can you share what you found with Excel? @:  | | |
|
| 19 |
| | | cynthia, 14:56 (15.06): Should I start a new textbox to answer questions? @: Message 11: To whole message | |
|
| 20 |
| | Bill, 14:50 (15.06): Use the text box to share your writeup. | | |
|
| 21 |
| | Bill, 14:51 (15.06): Use messages to respond to questions. | | |
|
| 22 |
| | Bill, 14:52 (15.06): Cynthia, is this yours? If so, can we erase it to make room for writeups? @:  | | |
|
| 23 |
| | | cynthia, 14:59 (15.06): Oh, well I was going to write up what I got using Excel, shall I not do that? | |
|
| 24 |
| | | | | Bethany, 14:56 (15.06): joins the room |
|
| 25 |
| | Bill, 14:53 (15.06): No, please do! And it will go nicely in your text box. @: Message 23: To whole message | | | |
|
| 26 |
| | | | | | Johann, 14:58 (15.06): joins the room |
|
| 27 |
| | | | | Bethany, 15:01 (15.06): Greetings everyone. I almost missed out, using the chat from yesterday wondering where everybody was! | |
|
| 28 |
| | Bill, 14:55 (15.06): Was this using Fathom's model tool or a scatterplot with regression line? @:  | | | | |
|
| 29 |
| | | | Beth, 15:04 (15.06): the least squares line from Fathom @:  | | |
|
| 30 |
| | | | Beth, 15:05 (15.06): with all the data | | |
|
| 31 |
| | | | | Bethany, 15:03 (15.06): Here is my response to the question | |
|
| 32 |
| | | cynthia, 15:04 (15.06): Now I'm having trouble typing in that textbox | | | |
|
| 33 |
| | | | | Bethany, 15:04 (15.06): hmm, that didn't work like I thought it would. @: Message 31: To whole message | |
|
| 34 |
| | | | | Bethany, 15:04 (15.06): I meant to point over to the white board | |
|
| 35 |
| | Bill, 14:58 (15.06): Looks like your figured it out! @: Message 33: To whole message | | | | |
|
| 36 |
| | Bill, 14:59 (15.06): To point, you have to first "draw" a reference rectangle with Reference tool. @: Message 34: To whole message | | | | |
|
| 37 |
| | | | | | Johann, 15:06 (15.06): Hi, Bill. I am here to help you with any technical questions about ConcertChat if you need @: Message 36: To whole message |
|
| 38 |
| | | | | Bethany, 15:06 (15.06): like so? @:  | |
|
| 39 |
| | Bill, 15:00 (15.06): Yup. @: Message 38: To whole message | | | | |
|
| 40 |
| | Bill, 15:00 (15.06): Excellent. I'll leave those to you, then. @: Message 37: To whole message | | | | |
|
| 41 |
| | | cynthia,
15:07 (15.06): Johann, I have a question. I can't get back into the
textbox that I started. Does it close or do I ned to do something
different? | | | |
|
| 42 |
| | | | | | | Gene, 15:05 (15.06): joins the room |
|
| 43 |
| | | | Beth,
15:10 (15.06): WOW. I like the Add Filter command. Bethany did you
paste the January months in another list? then teh same for the CO2
levels for January? | | | |
|
| 44 |
| | | | | | Johann, 15:08 (15.06): Which textbox is that, Cynthia. Can you create a reference to it? @: Message 41: To whole message | |
|
| 45 |
| | | cynthia, 15:08 (15.06): the "I am a novice" one | | | | |
|
| 46 |
| | | | | Bethany, 15:08 (15.06): No, I didn't paste into another list Beth. I just added the filter and graphed. | | |
|
| 47 |
| | Bill, 15:03 (15.06): Did you find that these varied from one to the next? @:  | | | | | |
|
| 48 |
| | | cynthia, 15:09 (15.06): Great thanks | | | | |
|
| 49 |
| | | | | | Johann,
15:09 (15.06): I see. I think it was just a matter of the background
color, Cynthia. Try it not that I made it green. @: Message 45: To whole message | |
|
| 50 |
| gerirose, 15:07 (15.06): leaves the room | | | | | | |
|
| 51 |
Bill, 15:05 (15.06): And, did you get a different answer? @:  | | | | | |
|
| 52 |
| | | | | | Gene,
15:11 (15.06): Hiya, Bill, and thanks for coming. Here's a photo of
Mona Loa. Does this excuse my not doing my homework? (I have a new
project with 4 Swarthmore students (you can imagine how that must be...) |
|
| 53 |
Bill, 15:06 (15.06): Is this really Mauna Loa? If so, the observatory must be one heck of a place? @:  | | | | | |
|
| 54 |
| Bill, 15:07 (15.06): Johann, Can we keep adding stuff at the bottom or do we have to erase? | | | | | |
|
| 55 |
| | | | | | Gene, 15:13 (15.06): Got it from the web, Bill, must be true. |
|
| 56 |
| Bill, 15:07 (15.06): Like the whales in Lake Michigan? @: Message 55: To whole message | | | | | |
|
| 57 |
| Bill,
15:09 (15.06): The next thing I would like to do is to have people make
screen captures of some of there Fathom work to share with us. But
we'll have to hope Johann can tell us how to get more whiteboard space. | | | | | |
|
| 58 |
| | | | | Johann,
15:16 (15.06): Well, there is a creative way of doing that. Bascially
we will have to rearrange some of the objects we have so as to make
space. Do you want me to do that? @: Message 57: more whiteboard space. | |
|
| 59 |
| Bill, 15:10 (15.06): Sure, give it a try. @: Message 58: To whole message | | | | | |
|
| 60 |
| | | | | Johann, 15:17 (15.06): O.K. Don't worry if you see stuff disappearing... it is all archived. | |
|
| 61 |
| | | | Bethany, 15:18 (15.06): Hey folks, I'll be taking an interlude, we have to do a fire drill. | | |
|
| 62 |
| Bill, 15:12 (15.06): Yikes! Come back when you can. @: Message 61: To whole message | | | | | |
|
| 63 |
| | | | | | | ihor, 15:17 (15.06): joins the room |
|
| 64 |
| Bill,
15:14 (15.06): OK, now that we've got some space, please arrange the
chat window and Fathom (or Excel) so you can see it. Use the screenshot
tool to capture something and paste it in (as Cynthia has already done). | | | | | | |
|
| 65 |
| | | Beth, 15:22 (15.06): How do you capture the entire fathom screen? | | | | |
|
| 66 |
| Bill, 15:14 (15.06): Don't try for the whole screen, just a portion of it. @: Message 65: To whole message | | | | | | |
|
| 67 |
|
| 68 |
| | | | | Johann,
15:20 (15.06): This is just my image from Excel, but everyone can take
snapshots of Fathom and paste theirs here for discussion. To create a
snapshot use the camera icon. You will have to have ConcertChat and
Fathom visible on your screen. @:  | | |
|
| 69 |
| | | | | | | | Annie, 15:20 (15.06): joins the room |
|
| 70 |
| | | Beth, 15:25 (15.06): I can not seem to paste my picture | | | | | |
|
| 71 |
| | | | | | | | | steve, 15:21 (15.06): joins the room |
|
| 72 |
| | | | | Johann,
15:24 (15.06): If you click on the camera, select a section of your
screen and then click on the whiteboard area your image should be
pasted. Anybody else having trouble with that? @: Message 70: To whole message | | | | |
|
| 73 |
Bill, 15:19 (15.06): Here is the graph that appears in the original data set. @:  | | | | | | | | |
|
| 74 |
| | | | | Johann, 15:25 (15.06): BTW, this displays only February Datapoints @: Message 73: To whole message &  | | | | |
|
| 75 |
Bill, 15:21 (15.06): This is a good question. Anybody have any thoughts? @:  | | | | | | | | |
|
| 76 |
| | | | | | | | Annie,
15:28 (15.06): Hi, folks. Sorry to be late. I won't paste in my whole
explanation, since I got a little longwinded. Basically I filtered the
graph so that it only included September, did a least squares line on
that data, then dragged to the appropriate point in 2005, and then 2004
(to match the available data). @:  | |
|
| 77 |
| | cynthia, 15:29 (15.06): Wjat | | | | | | | |
|
| 78 |
| | cynthia, 15:30 (15.06): What's the probability two people here have the same birth months Annie? I'm September too. :-) | | | | | | | |
|
| 79 |
| | | Beth, 15:33 (15.06): Help. can not capture my graph from fathom. | | | | | | |
|
| 80 |
| Bill,
15:25 (15.06): One way of looking at it is that the more of the data
your model accounts for, the more answers you can get from it. So,
while it's true you can get a prediction for a given month with the
filtered, data, the model with the sine allows you to predict for any
value, even half months. @: Message 75: To whole message | | | | | | | | |
|
| 81 |
| Bill, 15:26 (15.06): Does anyone have a graph that shows the sine wave plus the linear trend? | | | | | | | | |
|
| 82 |
| | | | | | | ihor, 15:33 (15.06): Cynthia, its about 10% :-) | | |
|
| 83 |
| | cynthia, 15:32 (15.06): Hi Ihor :-) | | | | | | | |
|
| 84 |
| | | | | | | | Annie, 15:33 (15.06): I sort of do....though my graph is scaled strangely. I'll get it. @: Message 81: To whole message | |
|
| 85 |
|
| 86 |
| | | | | Johann,
15:33 (15.06): Beth, do you have your two windows (Fathom and
ConcertChat) side by side? If you minimize ConcertChat or switch
applications it won't work. @: Message 79: To whole message | | | | |
|
| 87 |
| | | | | | | ihor, 15:35 (15.06): Am I impacting other peoples screens when I move things on the whiteboard? | | |
|
| 88 |
| Bill, 15:28 (15.06): Cool! Now, got a graph that shows the sine wave going through the data points? @: Message 84: To whole message | | | | | | | | |
|
| 89 |
| | | | | | | | Annie, 15:34 (15.06): So that just shows a very small slice of it, with the least squares line and my function. @:  | |
|
| 90 |
| | | | | | | | Annie, 15:34 (15.06): Well, sure, if I scroll to the left far enough. Hang on a sec. @: Message 88: To whole message | |
|
| 91 |
| Bill, 15:29 (15.06): Yup. @: Message 87: To whole message | | | | | | | | |
|
| 92 |
| | | | | Johann, 15:35 (15.06): Yes, the Whiteboard is shared, so everyone sees what everyone does. @: Message 91: To whole message | | | | |
|
| 93 |
| | | | | | | | Annie, 15:37 (15.06): So there it is. It's not perfect, but you get the idea. @: Message 88: To whole message | |
|
| 94 |
| Bill,
15:31 (15.06): About 5 people sent me writeups, and they were _very_
well done! But one interesting thing is that everyone's predictions
were single numbers. No one included a range. Why do you suppose that
was? | | | | | | | | |
|
| 95 |
| | | | | | | | Annie, 15:38 (15.06): My second prediction was a range. @: Message 94: To whole message | |
|
| 96 |
| Bill, 15:32 (15.06): Does it bother you that the points don't look like they are perfectly on the sine wave? @: Message 93: To whole message | | | | | | | | |
|
| 97 |
| | | | | | | | Annie,
15:38 (15.06): And I did that because I zoomed the graph in so far (the
first sine one I postd, without the data) that there simply was a large
range for 2004.7. @: Message 95: To whole message | |
|
| 98 |
| Bill, 15:32 (15.06): Sorry, I must have missed that. @: Message 95: To whole message | | | | | | | | |
|
| 99 |
| | | | | | | | Annie, 15:38 (15.06): Yes, but time was running out, so I decided I was close enough for now. @: Message 96: To whole message | |
|
| 100 |
| | | | | | | | Annie,
15:39 (15.06): There was a lot to read :-) I wrote, in my second part,
"I zoomed way in (see above) and found that the values for 2004.7 range
from 371.70 down to 370.95." @: Message 98: s | |
|
| 101 |
| Bill, 15:33 (15.06): I find these "wrinkles" in the data really fascinating. @: Message 99: To whole message | | | | | | | | |
|
| 102 |
| Bill, 15:34 (15.06): Uh, oh. Annie and Bill are dominating the conversation ... | | | | | | | | |
|
| 103 |
| | | | | | | | Annie,
15:40 (15.06): Presumably there is a way to get it to give me 2004.75,
instead of just .7. I didn't try to figure it out. The fact remained
that the prediction didn't get any better than the linear model. @: Message 100: To whole message | |
|
| 104 |
| | | | | | | ihor, 15:42 (15.06): Its fun watching you have fun, Bill | | |
|
| 105 |
| | cynthia, 15:40 (15.06): So why *should* the data fit the sine wave? | | | | | | | |
|
| 106 |
| | | | | | | | Annie, 15:41 (15.06): Gosh, that sounds like science instead of math :-) @: Message 105: To whole message | |
|
| 107 |
| | cynthia, 15:41 (15.06): perfectly | | | | | | | |
|
| 108 |
| | | | | | | | | steve,
15:42 (15.06): I guess I assumed that we were finding lines of best fit
and thus the predictions were assumed to be a range, for which the
number offered was a stand-in. |
|
| 109 |
| | | | | | | | | steve, 15:42 (15.06): In what situations is it worht making it explicit? |
|
| 110 |
| Bill,
15:37 (15.06): Challenge: Show us a graph using the data from Mauna2003
(Fathom sample document) that includes your model; i.e. superimpose
your equation on top of the larger set of data. | | | | | | | | |
|
| 111 |
| | | | | | | | | steve, 15:43 (15.06): and what would be learned in the process of trying to define the range |
|
| 112 |
| Bill, 15:38 (15.06): Basically, a single number is nearly useless because you have no idea how accurate it is. @: Message 109: To whole message | | | | | | | | |
|
| 113 |
| Bill, 15:38 (15.06): First question is, "Doe we have any _way_ to find it out given the data we have? @: Message 111: To whole message | | | | | | | | |
|
| 114 |
| | | | Bethany, 15:45 (15.06): Greetings I'm back! | | | | | |
|
| 115 |
| | | | Bethany, 15:45 (15.06): 11 flights of stairs later | | | | | |
|
| 116 |
| | | Beth,
15:48 (15.06): Could we find a Confidence interval on the slope of the
linear reg line then create two equations? I still can not get my
graphs to show. | | | | | | |
|
| 117 |
| | | | Bethany, 15:46 (15.06): What is a good way of determining the range for a prediction? | | | | | |
|
| 118 |
| Bill, 15:40 (15.06): Can you figure out how to do that in Fathom? @: Message 116: To whole message | | | | | | | | |
|
| 119 |
| | | | | Johann,
15:48 (15.06): Beth, another option for importing graphics is to take a
picture of your Fathom screen with either Grab (OS X) or Print Screen
(Windows) and use the "Add an Image" tool (before the camera icon) to
import it. @: Message 116: To whole message | | | | |
|
| 120 |
| Bill,
15:42 (15.06): Beth, after you draw a screenshot rectangle around
something, you have to click somewhere in the whiteboard. Have you done
that? @: Message 116: To whole message | | | | | | | | |
|
| 121 |
| Bill,
15:43 (15.06): Steve, I have an answer to the question you asked in
your email. about how to compare values from the Web data with those
given in the MaunaShort Fathom document. | | | | | | | | |
|
| 122 |
| | | | | | | | | steve, 15:51 (15.06): thanks. I am here @: Message 121: To whole message |
|
| 123 |
Bill, 15:45 (15.06): Here's the graph that shows both sets of points _very_ close together. @:  | | | | | | | | |
|
| 124 |
| Bill, 15:45 (15.06): Beth, since I have your Fathom document, should I capture something from it? | | | | | | | | |
|
| 125 |
| | | | | | | | | steve, 15:52 (15.06): so each point there is actually two points? |
|
| 126 |
Bill, 15:47 (15.06): Yes, here is a "blowup" of and area of the graph. @:  | | | | | | | | |
|
| 127 |
| | cynthia, 15:54 (15.06): Is there a way to share the question being discussed? | | | | | | | |
|
| 128 |
| | | | | | | | | steve, 15:54 (15.06): and this shows that the difference is consistent between the two? @: Message 126: To whole message |
|
| 129 |
| Bill, 15:48 (15.06): Which discussion? @: Message 127: To whole message | | | | | | | | |
|
| 130 |
| | | Beth, 15:56 (15.06): Thanks. I wanted to show the main graph with the entire data set. | | | | | | |
|
| 131 |
| Bill, 15:48 (15.06): Yes, it's an offset. I don't know why. @: Message 128: To whole message | | | | | | | | |
|
| 132 |
| | | | | | | | | steve,
15:54 (15.06): I think she means my question about why the values
differ between the data set in Fathom and on the web. @: Message 127: To whole message |
|
| 133 |
| | cynthia, 15:55 (15.06): Yes, that's right. Thanks for the interpretation, steve. | | | | | | | |
|
| 134 |
Bill, 15:49 (15.06): Here's is Beth's graph. @:  | | | | | | | | |
|
| 135 |
| | | Beth,
15:57 (15.06): I can not create a CI for slope, but I tried a inference
test for slope and Fathom found a t-value of 30.98. | | | | | | |
|
| 136 |
| | | Beth, 15:57 (15.06): thanks | | | | | | |
|
| 137 |
| | | | | | | | Annie, 15:56 (15.06): I just pasted a graph that meets your challenge...cool! @: Message 110: To whole message | |
|
| 138 |
| | | | | | | | Annie, 15:56 (15.06): This is it. @:  | |
|
| 139 |
| Bill, 15:50 (15.06): What's cool? What do you see in it? @: Message 137: To whole message | | | | | | | | |
|
| 140 |
| | | | | | | | Annie,
15:57 (15.06): That obviously we were given a small section of the
graph so that we'd think it might fit with a linear model. @: Message 139: To whole message | |
|
| 141 |
| Bill, 15:51 (15.06): I'm going to do a bit of cleanup so we can have space again. | | | | | | | | |
|
| 142 |
| Bill, 15:51 (15.06): No, I wasn't trying to manipulate your thoughts! :) @: Message 140: To whole message | | | | | | | | |
|
| 143 |
| | cynthia, 15:58 (15.06): Annie, would you explain more about that? | | | | | | | |
|
| 144 |
| | cynthia, 15:58 (15.06): Where is the graph. Could you link it? | | | | | | | |
|
| 145 |
| | | | | | | | Annie,
15:58 (15.06): We can always scroll back using the lefthand scroll bar
if we need to see old stuff on the whiteboard. It keeps a history. So
Bill, you oculd delete stuff, far as I know. It will remain in the
history. @: Message 141: To whole message | |
|
| 146 |
| | | | | | | | Annie, 15:59 (15.06): The graph is here, Cynthia. At least for now! @:  | |
|
| 147 |
| Bill, 15:53 (15.06): Sorry, I moved it. @: Message 146: raph is here, Cynthia. At least | | | | | | | | |
|
| 148 |
| | | | | | | | | | Craig, 15:57 (15.06): joins the room |
|
| 149 |
| Bill,
15:56 (15.06): We're about half way through the session. I propose we
take a moment to checkin with each other about how it's going. Could
everyone just write something about what this is like for you? Include
thoughts about the second hour. | | | | | | | | | |
|
| 150 |
| | | | | | | | Annie,
16:02 (15.06): ignoring the sine-nature of the graph, it looks more
like a chunk of a parabola than of a line, when you see all of this
data. @: Message 143: To whole message | | |
|
| 151 |
| | | | | | | | | | | David in DC, 16:00 (15.06): joins the room |
|
| 152 |
| | | | | | | ihor,
16:05 (15.06): I unfortunately dont have Fathom up and running. I'm
working from a spreadsheet. I assume CO2 function came from the plots
along the sine curve. Is that right? | | | | |
|
| 153 |
| | | | | | | | Annie,
16:05 (15.06): Ihor, I came up with that by plotting the function
a*sin(bx) + (the least squares lines), and messed with sliders for a
and b until it seemed right (and then figured out that b is 2pi...duh).
@: Message 152: To whole message | | | |
|
| 154 |
| | | Beth, 16:07 (15.06): I thought I was a good mulitasker, but aparently not. I just not started to follow everyone. | | | | | | | | |
|
| 155 |
| | | | | | | | Annie,
16:05 (15.06): You can plot a function on a graph. (We've almost
exhausted my Fathom knowledge, all gleaned from the resources Bill
suggested). @: Message 153: To whole message | | | |
|
| 156 |
| | | | | | | | | | Craig, 16:01 (15.06): leaves the room | |
|
| 157 |
| | | | | | | | | | David in DC, 16:06 (15.06): Hi All. I finally got everything installed have a few minutes to join in. |
|
| 158 |
|
| 159 |
| | | | | | | ihor, 16:08 (15.06): Annie, it appears to be a pretty good approximation. | | | |
|
| 160 |
| | cynthia, 16:07 (15.06): I'm enjoying trying to capture bits from all conversations... | | | | | | | | |
|
| 161 |
| | | | | | | ihor, 16:08 (15.06): I'm basing my birthday prediction on your function. :-) | | | |
|
| 162 |
| | | | | | | | Annie, 16:07 (15.06): It's quite nice for that one little chunk of the data we were given in the "short" document! @: Message 159: To whole message | | |
|
| 163 |
| Bill,
16:01 (15.06): For me as facilitator, it's pretty intense, but fun. The
hardest part is not being able to "see" people to gauge interest and
figure out where to go next. @: Message 149: To whole message | | | | | | | | | |
|
| 164 |
| | | | | | | ihor, 16:09 (15.06): Ihor smiles. | | | |
|
| 165 |
| | cynthia, 16:08 (15.06): you're doing great. roll on. | | | | | | | | |
|
| 166 |
| | | | | | | | Annie,
16:08 (15.06): (Disclaimer: Riz helped me with the math - I knew I
needed to add somethign to get it to "tilt", and he explained it was
the line of best fit.) @: Message 162: To whole message | | |
|
| 167 |
| | | | | | | | | | David in DC, 16:08 (15.06): The function appears more like a composition than a sum. |
|
| 168 |
| | | | | | | | | | David in DC, 16:09 (15.06): But, Annie's rotated sine curve looks good. |
|
| 169 |
| Bill, 16:03 (15.06): Why a composition? @: Message 167: To whole message | | | | | | | | | |
|
| 170 |
| | | | | | | | | | | Craig, 16:07 (15.06): joins the room |
|
| 171 |
Bill,
16:04 (15.06): Here is a simple regression model. Note that, along with
your predicted value, you get a confidence interval. @:  | | | | | | | | | | |
|
| 172 |
| | cynthia, 16:11 (15.06): so bill, you know we want you to show us how the technology does that | | | | | | | | | |
|
| 173 |
| | | | | | | | | | David
in DC, 16:11 (15.06): I would have tried to embed the sine inside the
linear transformation or taken the data and extrapolated it out far
enough by a transformation of the data. Sort of the brute force
approach. | |
|
| 174 |
| | | | | | | | Annie,
16:11 (15.06): Fair enough, but now you're venturing into territory for
which I need to learn more math, instead of just more fathom! @: Message 171: To whole message | | | |
|
| 175 |
| | | Beth, 16:14 (15.06): cool. Bill how did you do that? | | | | | | | | |
|
| 176 |
| Bill, 16:06 (15.06): Challenge: Show a residual plot of your model against the data. (Hint, see the Graph menu.) | | | | | | | | | | |
|
| 177 |
| Bill,
16:07 (15.06): The crystal ball tool in Fathom allows you to make
either simple regression or multiple regression models. @: Message 175: To whole message | | | | | | | | | | |
|
| 178 |
| | | | Bethany, 16:13 (15.06): crystal ball tool?? not familiar with that one @:  | | | | | | | |
|
| 179 |
| Bill, 16:07 (15.06): Cynthia, do you mean the underlying mathematics or the interface? @: Message 172: To whole message | | | | | | | | | | |
|
| 180 |
| Bill, 16:07 (15.06): Do you have Fathom 2 or Fathom 1? @: Message 178: To whole message | | | | | | | | | | |
|
| 181 |
| | cynthia, 16:14 (15.06): Fathom 2. | | | | | | | | | |
|
| 182 |
| | | | Bethany, 16:14 (15.06): fathom 2 | | | | | | | |
|
| 183 |
| | | | | | | | Annie,
16:15 (15.06): I assume that "your model" means I can't just do a
residual plot in the graph, and there's something else I need to do. @:
Message 176: To whole message | | | |
|
| 184 |
| Bill,
16:09 (15.06): Your model is your equation. A residual plot is
something you can get when you have a scatter plot and an equation. @: Message 183: To whole message | | | | | | | | | | |
|
| 185 |
| Bill, 16:10 (15.06): Then you should see a crystal ball in the tool shelf. @: Message 182: To whole message | | | | | | | | | | |
|
| 186 |
| | | | | | | | Annie, 16:16 (15.06): So if I just do residual plot on my graph, this is what I get. @:  | | | |
|
| 187 |
| | | | | | | | | | | Craig, 16:14 (15.06): leaves the room |
|
| 188 |
| | | | | | | | Annie, 16:17 (15.06): You mean the thing labeled "model", right? @: Message 185: To whole message | | |
|
| 189 |
| | | | | | | | Annie, 16:17 (15.06): (they wouldn't let you label it crystal ball, eh?) @: Message 188: To whole message | | |
|
| 190 |
| Bill,
16:11 (15.06): Now these are _very_ useful. When a model accounts for
all of the pattern in the data. the leftover, the residual, should just
be noise; i.e. normally distributed values. @: Message 186: To whole message | | | | | | | | | |
|
| 191 |
| Bill, 16:12 (15.06): Right. A model is kind of like a crystal ball because it allows you to make predictions. @: Message 188: To whole message | | | | | | | | | |
|
| 192 |
| | | | | | | | | | | Craig, 16:16 (15.06): joins the room |
|
| 193 |
Bill, 16:13 (15.06): Show us what the residuals look like for this one. I'll bet they are mostly noise. @:  | | | | | | | | | | |
|
| 194 |
|
| 195 |
| | | | Bethany, 16:19 (15.06): haha. I thought you were joking about the "crystal ball" | | | | | | | |
|
| 196 |
| Bill, 16:14 (15.06): What, me, attempt funniness? :) @: Message 195: To whole message | | | | | | | | | | |
|
| 197 |
| | | | | | | | Annie, 16:22 (15.06): So I've added that graph...not as useful, eh? @: Message 193: To whole message | | | |
|
| 198 |
| | | | | | | | Annie, 16:22 (15.06): there it is. @:  | | | |
|
| 199 |
Bill,
16:17 (15.06): I did it, too. Actually, it _is_ useful, since it shows
that for this portion of the data, linear plus sine does a _good_ job
of modeling. And in your graph, you can detect the non-sinusoidal part,
too, more easily than in the upper graph. @:  | | | | | | | | | | |
|
| 200 |
Bill, 16:18 (15.06): This is from Beth's solution, BTW. @:  | | | | | | | | | | |
|
| 201 |
| | | Beth, 16:27 (15.06): Did you send us the rest of the data after 1997? | | | | | | | | |
|
| 202 |
| | cynthia, 16:26 (15.06): What would the residuals look like if linear plus sine didn't do a good job? | | | | | | | | | |
|
| 203 |
| Bill, 16:20 (15.06): There's a Fathom document with it called MaunaLoa2003. @: Message 201: To whole message | | | | | | | | | | |
|
| 204 |
| | | | | | | | Annie, 16:27 (15.06): It's referenced in the tPoW that we worked with. @: Message 203: To whole message | | | |
|
| 205 |
| | cynthia, 16:27 (15.06): oh, i see they're all close to 0. | | | | | | | | | |
|
| 206 |
| | | | | | | | | | | Craig, 16:27 (15.06): You'd see a pattern... maybe larger residuals on one end, for example, or the residual graph curved @: Message 202: To whole message |
|
| 207 |
| Bill, 16:21 (15.06): Annie's first residuals plot shows residuals in the situation the data don't fit the model. @: Message 202: To whole message | | | | | | | | | | |
|
| 208 |
| | cynthia, 16:27 (15.06): thanks... | | | | | | | | | |
|
| 209 |
| | | | | | | | Annie, 16:28 (15.06): Beth's is more even than mine, meaning that her function is closer? @: Message 200: To whole message | | | |
|
| 210 |
| Bill,
16:22 (15.06): Annie mentioned that with all the data, it looks like a
parabola plus sine would be better than linear plus sine. Thoughts
about that? | | | | | | | | | | |
|
| 211 |
| | | Beth, 16:31 (15.06): What about exponential | | | | | | | | |
|
| 212 |
| Bill, 16:23 (15.06): Yes! And you could try adjusting the parameters to see the effect that has on the residuals. @: Message 209: To whole message | | | | | | | | | | |
|
| 213 |
| | | | | | | | Annie, 16:29 (15.06): Well, yea, it could be that too :-) @: Message 211: To whole message | | | |
|
| 214 |
| | | | | | | | Annie, 16:30 (15.06): I guess I'll go drag a slider! @: Message 212: To whole message | | | |
|
| 215 |
| | | | | | | | | | | Craig,
16:30 (15.06): Nature doesn't know parabolas from anything. The law in
Statistics appears to be, "if it feels good, do it." I would try
exponential (in fact, an exponential regression is easier to obtain
than a quadratic one). @: Message 210: parabola |
|
| 216 |
| | | Beth, 16:32 (15.06): Can I try to log the CO2s it see if it is an exponential. | | | | | | | | |
|
| 217 |
| | | | | | | | Annie,
16:31 (15.06): Oh, now mine is better - that helps a lot! Much like
moving a movable line while watching the sum of the areas of the
squares change. @: Message 212: To whole message | | | |
|
| 218 |
| Bill,
16:25 (15.06): I disagree. I think the law is to keep the model simple
except if you have some contextual reason to make it more complicated.
@: Message 215: The law in Statistics appears to be, "if it feels good, do it." | | | | | | | | | | |
|
| 219 |
| Bill, 16:26 (15.06): Speaking of sum of squares, did anyone try showing these with the linear plus sine equation selected? @: Message 217: To whole message | | | | | | | | | | |
|
| 220 |
| | | | | | | | | | David
in DC, 16:32 (15.06): Speaking of draging, I'm doing that to passing
NSF staffers to show them this discussion. They think it's very cool! | |
|
| 221 |
| | | | | | | | | | | Craig, 16:33 (15.06): @: Message 218: keep the model simple |
|
| 222 |
| | | | | | | | | | | Craig,
16:34 (15.06): I agree... my point was that if linear fits well enough,
use linear; if something else fits better, use something else. @: Message 221: To whole message |
|
| 223 |
Bill, 16:28 (15.06): Here's a portion of the graph you get. @:  | | | | | | | | | | |
|
| 224 |
Bill, 16:30 (15.06): Actually, this one's clearer. @:  | | | | | | | | | | |
|
| 225 |
| | | | | | | | | | | Craig,
16:38 (15.06): This graph is really cool. I was intrigued by something
you showed/said earlier about interval estimates for the birthday
thing. I'm new to Fathom; how does the "crystal ball" work? I say
(using the whiteboard history feature) a "models" window... how do you
get non-linear models included? |
|
| 226 |
| | | | Bethany,
16:38 (15.06): but I've heard that comparing different models is really
difficult. Isn't a function that contains more terms always have the
potential to fit better than something that doesn't? @: Message 222: To whole message | | | | | | | |
|
|
