search

UNCG-CSE / Library-Computer-Usage-Analysis

The University Libraries at UNCG currently track the state of a computer, determining whether or not a particular computer is in use. This data is compiled into a database, and a web app pulls from this database to show a map and number of available computers. As of Fall 2017, the data had not been used to determine which computers are used more frequently, aside from counting the number of times a computer transitions into/away from the 'in-use' state. This project attempts to correlate the usage of these computers with various factors, including: campus scheduling, equipment configuration, placement, population in the library, and area weather. Using this data, this project also uses machine learning to determine the best placement of computers for future allocation, and possible reconfiguration of equipment and space.
1 stars 1 forks source link

Presentation #54

Closed PatriciaTanzer closed 6 years ago

PatriciaTanzer commented 6 years ago

I'm going to start work on the .pwp document we already have this evening, ideas one what we should include?

PatriciaTanzer commented 6 years ago

One thing I'd like to do (we may not have time before the presentation), is to add some graphs to the UsageVSWeather file. I find the scatter plots of data vs lines of correlation a bit easier to see

brownworth commented 6 years ago

I will contribute more to this once I have tonight's homework done. I would like to generate a couple more graphs now that we have more comprehensive datasets. Some ideas:

PatriciaTanzer commented 6 years ago

Ok I ran into an issue with some of the graphs - The ones from UsageVSWeather don't include their x and y labels when copied. Do we want to use screenshots, or change the code in some way so that that works?

smindinvern commented 6 years ago

As ghetto as it is, a screenshot is probably the easiest fix, and should be sufficient for the presentation.

@PatriciaTanzer what do you mean regarding "scatter plots of data vs lines of correlation"?

PatriciaTanzer commented 6 years ago

I mean like in the example, where the data (such as gate counts per day) is a scatter plot, possibly two scatter plots if we are correlating two data sets (a correlation to gate counts might be hours of computer use per day), while the correlation between those data sets is graphed as a line going through the cloud of dots

In any case we should have one type of correlation graph (right now I think there are three) so that it's easier to see the comparisons between them

brownworth commented 6 years ago

I'm working on a couple of examples today and over the weekend. I had some trouble with how Bokeh handles datetime objects vs. the width of bars. I think I have it fixed. I'll post in here when I have more.

brownworth commented 6 years ago

I've sent all of you an email with a sample of what the visualization looks like at this point. Unfortunately, the GitHub interface doesn't preview Bokeh graphs well (if at all).

PatriciaTanzer commented 6 years ago

I've added screenshots of the utilization graphs that Nick did to the powerpoint. Can you each take a look and make sure that what you want to talk about is in there? I grabbed what I thought was most important, but that might not be the same for everyone.

Also: Should we link in the bokeh graph that Brown's been working on or wait until the next presentation?

smindinvern commented 6 years ago

@PatriciaTanzer: Thanks! I actually just updated the UsageVsWeather.ipynb notebook to include the full data set, which rather significantly changes the plots, so I'd like to show a before and after comparison.

I don't have MS Office, so I'm not sure if I should directly change the powerpoint. If I push a modified version to the nick branch could you guys just make sure that it renders correctly in Office?

Regarding the bokeh plot that @brownworth mentioned, first off: good job getting up and running with bokeh. I think it's worth including in the presentation even if just to show where we're going with the visualization part of this assignment.

If possible, I'd suggest reworking the plot a little bit, as currently I think the presentation is a tad confusing. As @PatriciaTanzer already mentioned, axis labels would help a lot with that. Also can the y-axis values be formatted as times? e.g. 12am, 1am, ..., 11pm. Also, changing the x-axis values from range(0,7488) to e.g. list(itertools.chain.from_iterable([[x]*24 for x in range(0, 312)])) would mean make the plot a true 24x312 grid of points, and should make the x-axis more directly meaningful.

Like I said, either way I'd vote that it be included.

mtellis2 commented 6 years ago

I say we include the bokeh plot even if it's not a finished product yet, like Nick said it helps to show what we're trying to look at and how we're trying to visualize it. But labels would help to understand the plot a little easier. I'm going to take a look at the updates you made @smindinvern and I'll check to see how it looks in Office.

smindinvern commented 6 years ago

@mtellis2: thanks. I just pushed the new ppt to nick. permalink

mtellis2 commented 6 years ago

@smindinvern the graphs on the ppt all look the same as they do on the ipynb.

smindinvern commented 6 years ago

ok, i merged in my changes. hopefully everything looks ok.

PatriciaTanzer commented 6 years ago

Looks pretty good. I'll fix a few things - The title on the first slide went missing, and some other info got dropped, but that's easy to redo

PatriciaTanzer commented 6 years ago

Lol the title didn't go missing, it was turned into white text on white background

brownworth commented 6 years ago

RE: axes, I just removed the x-axis, but the mouseover tooltips now show computer name, so I feel like that's a better way to convey what we're showing. Let me see what I can do to make the x-axis work. Unfortunately, I can't seem to pass a 24x312 dataframe, with cell values as a 3rd dimension to Bokeh. It wants single dimension arrays for each.

brownworth commented 6 years ago

Unfortunately, the index along the x-axis needs to be discrete values. image zooming in: image

brownworth commented 6 years ago

Well, good news and bad news: zoomed in, it looks like this: image

but zoomed out, it looks like this: image

brownworth commented 6 years ago

So, I'm going to hide the x-axis for now. Unless I figure out something else in the next 4 hours.

PatriciaTanzer commented 6 years ago

Hey just noticed we now have a ppt in docs AND in the main folder - I thought we were using the one in docs?

mtellis2 commented 6 years ago

@PatriciaTanzer hey I think this was my fault, in “develop” I only saw the ppt in the main folder and that’s where I made the updates.

brownworth commented 6 years ago

I uploaded the ppt and html directly to the docs folder in master just before class. Those were the ones we used yesterday.

On Mon, Oct 30, 2017 at 1:53 PM, Michael Ellis notifications@github.com wrote:

@PatriciaTanzer https://github.com/patriciatanzer hey I think this was my fault, in “develop” I only saw the ppt in the main folder and that’s where I made the updates.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/UNCG-CSE/Library-Computer-Usage-Analysis/issues/54#issuecomment-340529550, or mute the thread https://github.com/notifications/unsubscribe-auth/ASkUrSsknY2UwHKWc7tYiZDqhHbajvwGks5sxg01gaJpZM4QGfXF .

PatriciaTanzer commented 6 years ago

We've finished this weeks ago