Tip:
Highlight text to annotate it
X
Event ID: 2103291
Event Started: 3/28/2013 3:00:00 PM ----------
Please stand by for realtime captions. >> Hello everyone but we are going to get
started in just a minute but I just want to let you know if you are having technical difficulties
you can call one 800 263 Hello everyone but we are going to get started in just a minute
but I just want to let you know if you are having technical difficulties you can call
Thank you and we will get started in 2 min. option one. Thank you and we will get started
in 2 min. >> Hello everyone and welcome to the webinar.
I am [ Indiscernible ] the program manager digital of University. And we have a great
event planned for today on data visualization. >> We have invited one of the agency leaders
in this field to talk about how turning data into representation can help mitigate that
data and provide meaning and understanding at a glance.
>> Let me introduce the presentation of the Census Bureau -- Eric Newberger -- is the
assistant -- Eric Newburger -- he is the assistant to the associate director of communication
and one of the leads of the Census Bureau's efforts to use the date hour of data visualization
to open its data to a broader public. >> We also have Mark -- Marc Perry who is
the chief of the relation branch of the Census Bureau's population division. And the technical
lead for the bureaus dated the -- date it this relation effort.
>> And finally Alex Tait is the chief photographer and vice president of international mapping
in Maryland. As a contractor with the populations divisions, Alec and his team assist in producing
the data visualization of the week. >> And here is Eric.
>> Thank you very much. And thank you very much for having us. We are all three of us
excited to be having this discussion. >> We have the data -- the Census Bureau's
data visualization mission. We have the program and the intent is to greet this increase the
ratio graphics to text to the Census Bureau publications both online and in print, and
to open the databases and analyses to a broader public.
>> How to get to the point where we needed a program to do that?
>> We have -- history of the Census Bureau going back sometime -- the map that you're
looking back is from 1861 actually. And is now was one of the first [ Indiscernible ] not
ever put out. The Orthodox max that I know of was invented 20 years earlier by some officials.
This is the first one that I am aware of that was put out by a government agency.
>> This particular map is the slave population in the southern states. This map actually
have a considerable historical importance as you might imagine to the subject matter.
But step back from the main -- a moment and consider this was the census bureau and/or
almost the sheep are always the first time in the world, we were resenting people -- presenting
people as geography. That was not the time. >> In 1870 to 1890, and even into the early
1900s we were producing atlases that were filled with small multiples that tell people
[ Indiscernible ] and that were filled with pages that look like this. Pages that were
covered in the presentation of data in a visual way. To open up these dated to a broad audience.
And then what happened? >> Well the [ Indiscernible ] tabulating machine
happened. In 1890 for the first of the census had a machine I could do tabulation. Than
this machine was a resounding excess in 1880 -- said this. So eight years it took to process
the 1880 census took I believe a little over two years -- and we actually did it twice
so that we could check the results. >> Now that idea that we had the chemical
aids for tabulation was that led to an explosion in the availability of data from the Census
Bureau. Moving into the 20th century. Women to electronic disease -- electronic machines
and him actually the large-scale machine that we have in the publications all started to
correct this or this. >> Or this. Because what we have is we are
tabulating machines. That was the stated the art -- the state-of-the-art was a well organized
well cut table data processed in a way that could be useful provided you are willing to
redo the table. >> And reading through the table was really
necessary. We were putting out whole books of tables and for the people, the researchers
who could spend the time and had the energy and the know-how to go through those tables
-- we were providing data that had never been available before in the United States or most
any other nation. >> And it was a real new. But -- boom Huck
>> But at the same time I was lamenting the audience. It was limiting the audience to
people who could see the table with very rare exceptions. We did on occasion put out graphics
with -- like what you are looking at now which I think is one of the great graphics of all
time because one.is 7500 people. So we're actually seeing the people of the United States
come together to make the United States. There is actually not picture of the United States
[ Indiscernible ] it is all just data. >> And so I love this one but it was very
rare. Very very few images that you could find especially during the latter half of
the 20th century in the Census Bureau public -- publication relative to the number of people
the table. >> And of course the ultimate version is the
American factfinder. American factfinder is the data submission tool that the data sent
-- census bureau is currently using. Amy not know the applicant is. Is approximately -- at
last that I am aware of -- 370 -- at last that I am aware of -- 370 billion table cells
inside American factfinder. 370 inside American factfinder. 370 billion. Most of them have
[ Indiscernible ] associated with them so you can actually pretty much double that number.
>> Is an enormous data sent. That is the thing -- we are really not talking bout tables anymore.
We have gone beyond an analytical table. >> We are talking about data sets.
>> You have to read through to understand it to make sense of it.
>> We want to open up American factfinder. We want to open up all the tables in the Census
Bureau. We want to open up the data set that we have for that broader public by providing
visualization that can open up those data sets to people who don't use tables.
>> And even though we have been making improvements for American factfinder that you see on the
screen among the indexing improvement that we have been making still -- it still ultimately
is about finding your table. Like how do you envision what is in the table.
>> So this is that Schneider -- Schneiderman. Reading -- readings in information visualization.
In regards to interactive visualization. >> Then Schneiderman and is up at the University
of Maryland have been instrumental in -- okay -- just if you look at that smart phone in
your pocket, a lot of the interactions you are having with it like your sightings to
open the screen and other such -- that came out of his lap. That is his work. He invented
[ Indiscernible ]. He did a lot of stuff pocky change -- the talk that he came to talk to
us at the Census Bureau. Roughly about one a month -- to talk about the work that they're
doing. When I tell you that we are 100 years in the Census Bureau, we would just producing
tables. We were producing analysts who thought in tables. We were producing analysts who
knew how to make tables. Really really good tables. But what we're doing is making tables.
>> To ship to visualization. To open up the data sets for visible -- visualization requires
us to change the culture with the building. And that is what the data visualization project
Census Bureau is really about. >> And it starts with bringing in luminaries
from the field to talk to us about visualization. What is going on in the world so we can get
excited. And learn a little something at the same time.
>> We brought in the panel of -- from the Washington Post. Their graphics people. We
brought in someone from the New York Times. We brought in Stephen few. Is a corporate
data trainer in Basle edition pocky was but one of our monthly speakers and fill the -- filled
the auditorium. But also you say the first couple of days to do the first of our formal
trainings in addition to the monthly visits with formal training.
>> Naomi Robbins published creating more effective graphs. This is the guy but to creating graphs.
With academic work. And she is our -- one of the core instructors for the ongoing classes
that we are doing on the data business. >> We have other sectors of the people coming
in. I don't want to show you a list but I just wanted to give you a sense of the sort
of the effort -- is really about the minds of the analysts more than it is about any
particular tool or software or anything else. >> One think everything on the line --
>> We have a data visualization team of which 63 members -- with 4000 people is a huge number
but 23 different areas are contributing people. >> Each one of these -- of these are all these
are the prime areas within the Census Bureau and the publication oxo that we have someone
or -- someone from everywhere or a couple of people from everywhere. To explain how
we can build the auditorium we have the event. >> Because overly Is a network that extends
throughout the building. We have an internal website to support this network.
>> And we also have devastation projects. Of the devastation projects were intended
-- intended to provide a prime location on the front page of the website. To give prominence
to data Islamization as well as showing the people within our building to take people
that existed in a different context and review different -- review different things.
>> Immunities to -- devastation projects as the visualization of the week.
>> How internally we have requirements for the data visualization of the week -- four
or more dimensions of the analyzes whether data or an attempted -- annotative -- and
there is a list of these and each of these have subheadings and we don't even need to
know about all of that. >> Let's actually go to the picture. Alright.
So here is one of the most popular data visualizations that we have. Up until last week -- which
I imagine we will come up with something -- this is eyelids of high income. Now the idea here
is to just take the ACS meeting household income by county data. -- Median household
income by county. >> And provide a very minimal interaction
to see what counties are coming up -- and the interaction is just as later across the
bottom of the screen at present the cider is all the way to the left. So any County
that has a median household income higher than $18,000 shows up in green. And of course
everything shows up in green. >> But what happens if we move it later over
$10,000. Now we understand why it is called islands of high income. Because we see these
isolated counties. >> Or in the Midwest. We see the Northeast
four-door as this long stretch of high to medium household income or county. This continued
was patterned. Minute islands of much as a a cappella.
>> We see how on the West Coast we see the California and Los Angeles areas -- the Chicago
area of the lakes. We see these different areas and we have an understanding now of
the wealth distribution in the United States. Which all of these data were previously available
in tabular format through American factfinder. You can download that. If you greatest picture
-- or you can come to this -- the date of this relation of the we can use the slider
and in about 10 seconds get this kind of understanding. >> And you can go the other way also.
>> Move the slider down again so now you're looking anything is greater than the $2000.
Which is pretty low compared to median value of around [ Indiscernible ] thousand.
>> $32,000 or more shows up in green. Which means anybody who is in white is less than
$32,000 as a median household income in the County
>> We see a different grouping. A different pattern emerging.
>> To this one simple interaction, this one -- pretty much the simplest possible interaction
you can have, turns what was a data set that was a table of 3141 colonies -- and it turned
that table into an understandable analytical tool that really does open up the data set
to a wide public. >> So that's what we want to do -- that's
what we want visualization to doing this: what actually all of our Islamization to do
is used to strive to become >> And this one doesn't.
>> Enough to introduce the person who created this visualization I would like to introduce
Marc Perry who is my partner and data visualization here at the Census Bureau.
>> Great, thank you Eric. This is Mark. So I just wanted to let everyone know about some
upcoming enhancements in the functionality of the visualization of the week that we will
be rolling out over the next probably 3 to 5 weeks or so
>> So for people who have been familiar with the visualization of the week Gallery -- certainly
one of the things that you have noticed is we have not really made it easy as of now
for people to contact us -- there is in fact no phone number, no dedicated e-mail address
-- no common feature -- there's really no easy way for the general public to communicate
with us about a particular visualization. >> And a people have based on their own creative
ways -- people know somebody at the Census Bureau and say hey -- can you get this message
to the folks who did this. >> Well we're entertained all that.
>> To the coming weeks we will be adding some nationality. We will be adding a comment feature
for each visualization. So you will be able to -- there will be a little icon on the top
right corner of the visualization. You will be able to click that to provide any comments
or feedback. >> There will be a little thumbs-up/thumbs
down rating. For each visualization that you will be able to use. And also have a dedicated
e-mail address. So if you have suggestions on future topics, if you have questions or
need clarification. >> We will finally be making it easier for
people to communicate directly with us. So if you have questions about how do you do
this, or where is the data set or really any communications of any kind, we will essentially
just be making it a lot easier for that to happen.
>> So now I would just turn things over to Alex and he can talk about the details and
sort of how we make all this happen. >> Alex, go ahead.
>> We cannot hear you Alex. >>[ Pause ]
>> Alex -- I think you unmuted box Mac Alex, on your audio control panel, unmute your microphone.
>>[ Pause ] >> Can you
hear me now ask --? >> We can
>> Sorry about that Oaks. -- Folks. >> My name is Alex Tait and I work at a company
got international mapping. And we are a contractor with the US Census Bureau and we have been
working with Eric and with Mark on the data visible addition of the week. And I'm here
to show you a little bit about the nuts and bolts about how we have put together these
data visualizations together. >> These data visualizations are in the gallery
that Mark was talking about. Here is the web link if anybody wants to go and play with
some while I am talking. And I urge you to do so.
>> Before I get into the how-to, I wanted to talk a little bit about the parameters
within what we are working for this visualization. >> We are of course working with the Census
Bureau data. And as Eric was pointing out we are dealing with these visualizations for
the general audience. The ideal is to show people of the United States with the sense
data looks like and how they can see different patterns.
>> We limit ourselves to a canvas item about 880 eight 80 x 6 60 pixels. And these last
two are very important -- we wanted the data visitations to be compatible with as many
browsers and as many platforms as possible. So these website based visualizations are
compatible back to Internet Explorer seven and they are fully compatible with Apple IOS
devices. And of course anybody who has been working in digital media knows that that has
some implications such as the fact that there is going to be no Adobe flash. Which means
we are doing primarily our interactivity with JavaScript and HTML and lots of images.
>> We had to learn to love images -- so there is a lot of the interactivity that you can
see that looks like animation is actually very quick image swapping talk -- and a lot
of the interactivity is fairly simple. As Eric pointed out we are going for simple and
effective on the interactivity so that we have cross-platform compatibility.
>> So this all serves to keep the focus on good visual ideas and not cutting-edge interactivity
or animation. >> So how do we put one of these together?
We are going to take a look at the blooming state visualization and I am going to discuss
seven stages of the process. And talk about the tools that we use to accomplish each of
these stages. >> Let's take a quick look at the booming
sates visualization. I will go to my browser and scroll down to blooming states. And you
can see a map of the United States with these sort of what we call the -- we call this blooming
stage because we believe like flowers. But each of the circular graphs is representing
population change. In each decade from 1790 until 2000 for that state.
>> So if we take a quick look at New Jersey, you can see that the radio rafts starts at
12 clock in the 1790s. And goes around for each of the decades old weight to the 2000.
At around 11 o'clock. >> And we can see -- all the way to the 2000.
>> The populist change has been a positive change because the warm colors have been the
population changes -- so there's been a decrease in the population. You can see how fast the
jersey has grown. Is growing faster than 18 1830s and 1840s -- and then there has been
a gradual slower rate of change. >> So loudly with something like this together?
The first eight -- so how do we put something like this together? >> First subject the concept. How do we put
this -- the tools here are putting together your thinking and your brainstorming and starting
out with either questions about data, or data set that you want to look at and try to develop
questions out of. >> I had tossed into the mix the idea of bringing
images so you have some idea that maybe you want to show up with a graphic, you are to
show it with a map. Pencil and paper come in handy. Voices -- is generally a group to
be involved in putting together a concept. And important to have a discussion. And so
we have the previous data visualization that we worked on, but also the experience of a
demographer or Jennifer in looking at the data.
>> So the question we had for blooming state is what does decade to decade change in population
look like for all the states. And here's a quick visualization -- of course I am in the
concept stage but I have already shown a rough wrap if you want to take a quick look at some
the population changes by decade for all the different states.
>> And this is one quick look at it. >> So I will be looking at the stages but
of course socially early on in the project -- process, your concept stage is David Miller
-- motivation stage in the rough graphic stages all intermingle as you trying to figure out
the best way to show what you want to show. >> So we refine the question down to how to
wish a regional differences in state population growth?
>> What I'm calling stage II and again the stages do intermingle with the data preparation
our primary tools are Microsoft Excel and ArcGIS for databases.
>> So many of you know that this is a wonderful tool for manipulative data. And has many good
tools for is alleging data in reference format. >> Those of the primary tools.
>> This second site about data shows that in the background you can see the raw numbers
of population. We then manipulated the raw numbers to look at the percentage of change.
So you can see that for New York in 1790, there was a 73 -- in 1790s that is -- from
17 9217 from 1792 1700 -- there was a 73% increase.
>> Third stage will be the rough graphics -- the primary tools here is a sad -- micro
soft Excel is very useful. ArcGIS For preliminary mapping. And this is where we bring in Adobe
Illustrator to do the graphic thinking -- is a canvas for us to using graphic taking. So
often we are using elements out of Excel and ArcGIS and bringing them into Adobe Illustrator.
>> In looking at decade-old population change for each state, we wanted to take a look both
at the percent of maximum population, and at the percentage change. You can see that
for most states and certainly for Iowa, there is a radical visual difference between these
two ways of looking at the data. >> So for Iowa on the left, you're looking
at in each decade how much of a percent of the maximum population, which for Iowa was
in 2010, was the population [ Indiscernible ] [ Indiscernible - low volume ]
>> You can see that I have again published very quickly. And each near maximum population
as he reached maximum population very early on.
>> The percentage change graph shows that in a different way. So that you can see the
radical percentage change in about 1840 when Iowa first started growing.
>> So we have to make a decision and further refined the question and what we agreed to
do the data we decided we wanted to take a look at the percentage change.
>> That brings up how that would want to show it. And we thought let's get people thinking
a little bit differently. And take that our graph and wrap around a circle so that we
will be able to start looking at the data in a different way. And you can see here the
transformation from a regular bar graph to a radial bar graph. Anything for you appreciate
more help that changes your way of looking at the data, is when you see it in looking
at the entire United States. >> So even in thumbnail format here for the
storyboard but you can see that the flowers in the Midwest look very different from the
flowers in the West >> Or in the Northeast. That is a regional
pattern that we wanted to show. >> So stage IV storyboard -- this is very
important for the interactivity thinking about the interactivity -- we just lay out a simple
series of -- thumbnails to show what would happen if the user is manipulating this data
visualization. >> In this case is a simple mouse over -- one
of the -- you have an initial map. You have a mouse over one of the states. Any pop that
it pops up a detailed graph. And this allows you to see both at a national level and then
at a state level, different layers of detail about the data set.
>> So once we have established the storyboard, and once we have established the way that
we will show the data come in that we are able to move on to the final graphics -- in
Adobe Illustrator and Adobe Photoshop and we have primary put images for quickly loading.
And you can see some of the images that we use for one of the states here.
>> And then we move on to the prodevelopment -- codevelopment stage. Primary tools here
where the Adobe Dreamweaver or you one you the way of coding JavaScript in HTML. An excellent
appointment that we use the Adobe extended script toolkit to do some special work in
Adobe Illustrator. Where we created our -- our developer created a way of generating the
circular bar graphs -- the radio or graphs automatically from a data set.
>> Which I hand it would have been very difficult. In the show some of the crowd -- code that
he created for some of the radio bar graph. >> Programmers are quite useful for writing
shortcuts for production >> And then lastly, the web presentation is
putting everything together. Using Dreamweaver and other tools for putting together HTML
pages. And part of that process is testing in the standard browsers. The explores -- Internet
Explorer, Safari Firefox and chrome. So some of these get you mingled, especially in the
early stages >> I thought it would be useful to highlight
some of the tips that we have learned. In working with Eric and Mark and working with
a programmer and the other folks at senses that are involved in this project.
>> From a concept side of things, as a mentioned it is very important to discuss the concept
out loud with others. And I would say that it is a *** to note that those people on
your team, but also people not under team. People that are not directly involved. We
have got a lot of good ideas on feedback from friends and family. So I encourage you in
coming up with concepts, to discuss these things with people who are not necessarily
experts in demography or in data visualization. >> Gather lots of visual ideas -- and we discussed
the ways of looking at things. Include your developer and programmer in the early stages.
Oftentimes the developer is only brought it at the end after things have been thought
through. And people that know how to code and know how to put together interactive elements
are very useful to have at all stages of the process.
>> And lastly, shortcuts -- don't shortcut to brainstorming. You to generate a lot of
ideas because as I will note later they will not all be the best ideas so you need a lot.
>> Underproduction -- note that it is very easy to copy and paste from Excel to illustrator.
So if you are good at visualization -- this was by seeing graph -- if you're good at visualizing,
you cannot just copy of the state graph from Excel to illustrator. You can create simple
graphs directly in illustrator that I live. Connected to data sets. Encourage your developer
to create a library of widgets. Agency as you look through the gallery that there are
certain interactive elements that are used on multiple pieces. And that saves a lot of
the time when you are looking -- coming up with data visualization -- every week and
the clock is ticking >> Lastly -- be careful. Don't review digital
projects by -- products by printing them out. It is important to review digital products
in the media by which they are intended to have a better sense of color and type size
and graphics effects. >> Lastly it's about process. Be sure you
have an iterative creative process. We work closely with Mark and Eric to make sure that
there is multiple pages of review. You need to be able to revise you need to have a lot
of people looking at things so that you get good ideas about how to improve both the interactivity
if there is some, and the visual oppression of your graphics.
>> And lastly, generate lots of visual ideas because you should probably kill at least
half of them. We have a virtual floor littered with dead ideas that start off as something
that was promising and we probably found a better way to do it but there were probably
three or four ways that did not work well. >> Alright -- enough about the process. I
would like to show you a couple three samples from the gallery.
>> So wickedly blooming states. And I will score down to the very first data visitation
that we created. >> This one is called top 20 cities. And it
is a word map, a literal word map. With the names of cities -- are generally in the correct
geographic location. And they show how many of the senses decades from 1790s two 2010
-- have a time that city was in the top 20 state
>> And so you can see the large East Coast cities -- Washington and Baltimore, Bill Duffy,
New York and Boston -- they have always been in the top 20. So if you take a look at Baltimore
and you can pop up a graph and you see its position in the top 20 is ranked fifth up
to second, down to 17. But always in the top 20.
>> Bus people don't know that Baltimore was the second-largest city in the US for three
decades. >> We can take a look at New York. Number
one. >> We can take a look at New York. Number
14 every census ever from 1792 2010. >> Now an unusual pattern that you see here
is the city like Los Angeles that many people consider a very important city but of course
it is smaller on the map because it only contains one of the 20 -- largest in 2010.
>> You start seeing a different pattern -- and this is a historical pattern that you're looking
at. Touched a contemporary pattern. So another level of looking at the data.
>> One of our popular data visualizations, was when we created initially in print. And
that you need to take this data and show it interactively in the data visualization for
the website. So we were looking at population density along interstate Interstate 95.
>> And we wanted to sort of tend you're in a car driving along Interstate 95 -- have
been so the population. And as Eric has said this sort of data is available on the track
level or county level on tables for the Census Bureau. But it has been sort of were difficult
to look at and essentially more difficult to look at in table form when you're thinking
about a profile along the road. >> So I will go ahead and play this. And you
can see as the car travels up I-95, you could see the major cities. There is the megalopolis
from Washington to Boston and you can see the cities along I-95 as increases in population
density. >> Again for those of you looking at the technical
side of this come up this is a very quick image dropping out and not a true information.
And this works in Internet Explorer seven in machines just fine.
>> I'm going to return to the gallery and show you one final data visualization. This
is one that came out a week ago. And I think this is the most popular one to date. And
who have not had a chance to play this game I certainly urge you to do it. This looks
very much like the March madness brackets that you would probably fill out in your office
pool. But in this case, we have substituted the team with metro areas. And we have matched
up -- very different metro areas in the United States. And your goal of the game is to pick
which one is greater. So I have Portland and Buffalo here. And I will guess that Buffalo
and Portland -- [ Indiscernible ] so when I find I wrong come up or Louisiana Buffalo
is a red. I have Rochester and Bakersfield. I get that one right, it turns green.
>> And you can see of course that some of them are easy and some are hard.
>> And so Los Angeles goes all the way up. >> This was a way to connect both the popular
culture phenomenon and also this was timed for the release of the new estimates for the
metro area population. >> It was way to tie senses point -- Census
Bureau even to popular culture even to put into debarment that did very well with people
commenting on it. And we were written up in the Hartford current as the lovable nerds
at the Census Bureau. >> So that's it for my presentation and I
will pass it back to the webinar coordinator's. >> Thank you Eric, Mark and Alex. We're going
to take questions. But before we do on the digit is a special guest, Jean Holmes -- is
the digit to go -- specialist. And just talking about the communities, open visualization.
>> Thank you so much. >> I am coming through?
>> Yes. >> Great.
>> Will like you guys to know about data.gov is that we have a lot of visualization capabilities.
As well as 365,000 geospatial data sets. >> The data.gov is an open government initiative
from the US office of [ Indiscernible ]. And we provide access to over 400,000 data sets
from 185 different federal agencies. >> We also go beyond the federal agencies
to cities, states and counties. As well as international partners like the United Nations
and [ Indiscernible ] >> You can finalize the data there and you're
able to do a variety of positions on it. >> So the old Geo spatial -- topics in
>> -- Is actually partial now at the deck of.
>> You can do mapping, visualization. You can also do visitation a little bit as Alex
was talking about -- you can make charts. And interact with maps. Looking at matching
up different kinds of data sets together from different agencies. Said Mac also on data.gov
we have a variety of community. These are topical areas international ours interests
-- they range from things like energy and education to business an infection ring out
and some of these communities are very visually focused with data focused.
>> The oceans community for example focuses entirely on coastal and Marine special planners.
So these are folks are really looking at mapping data and to look at the addictive analysis.
And the use of coastal areas and around the lakes.
>> So there are a variety of resources -- that we have a data.gov and access a lot of different
data. So we would love to have you guys come in and join the community. Look at the data
and create visualization that you can venture backed. To the public. Suggests new data sets
that you would like to have access to the federal government. And we will work her to
make this data sets available to you. >> Great. Thank you so much Jean.
>> So let's jump into the Q&A. I have a few questions here. And Eric I think we go to
you. About by the weight accessibility. To talk a little bit about what kind of process
to go through to make sure that the data visitations are -- that meet by the late -- law and also
one attendee engine that of course there is [ Indiscernible ] but what else could folks
do to make sure that the graphics made by the lead standard.
>> There are -- okay there are three things you have to know. And if you're asking the
question you may already know some of that. What everybody knows that color choices. Colorblindness
is not actually colorblindness -- you're not blind to color -- their particular fields
of color that you don't see. And their different ones depending on the particular type of impairment.
>> So choosing the color palette which has a lot of contrast. But does not dazzle your
eye if you have normal vision that is sort of high analyst. Now there are services that
we have been looking into that will actually -- that are online of their free. And you
can actually submit your visualization to these services and they will send you back
what it looks like to someone who has a particular type of color impairment.
>> And that could be helpful. But just picking a regular color palette that is very very
handy. >> One of the part about 508 compliance sexually
the interactivity itself. Write properly speaking, ultimately you should be able to do things
with your keyboard instead of the mouse. Because they're people that cannot use a mouse.
>> That is part of 508 compliance. And that's the one that most of the think about. If you
can have keyboard controls in addition to the house, that helps a lot.
>> And the last thing is -- and Mrs. the big fallback position.
>> Is -- and this is true for all of the Viz-of-the-Week and productivity us want to do. The data needs
to be available in the tabular format that is machine readable. And you need to be able
to contact from the visualization. So if someone says I can't see this visualization, I have
no ability to work with it. >> The data need to be available in just a
regular table. So the table behind the visualization needs to be available with one click. And
it needs to be available in one of the standard formats that the machine reader could read.
So theoretically someone could say -- have a listening device or a machine reading device
that they could listen to that would read out: title, road title -- etc.
>> And those are the three things to know about Bible week plan Phillies are my perspective.
Mark or Alex if they want to jump in and would be happy to hear them.
>> [ Indiscernible - multiple speakers ] >> I think you hit it just right.
>> Okay great. The next question -- the census flow maps as a uses flash -- is used as exception
to what you mentioned Alex about using image swapping?
>> There are a few exceptions. And to gain additional interactivity, there are a few
exceptions. >> In general we try to make sure that everything
is really cross-platform but that was one case where to get the additional interactivity,
the flash environment was limited. >> Okay great. And can you identify what kind
of files that you use to export from the [ Indiscernible ] to take into illustrator.
>> Sure. So sometimes we are coming out of ArcGIS straight to JP or JP e.g. files and
actually skip the Adobe Illustrator and take those images right into Photoshop. But if
we are going into Adobe Illustrator, there is an export as adobe illustrator format.
>> To come out of our -- to come out of -- so if you come out with another component out
of a RC map ArcGIS or -- you can come out of that
>> One of the key things to do is make sure the resolution on the export is very high.
That to 1200 or maybe 24 under -- so that you get a level of detail that will produce
a satisfactory graphic image. >> Okay great. Specifically on the radio charts,
is any convention about what point should be at the total cost position -- the 12 o'clock
position? >> I think the convention to follow is that
the start point that you want people to initial -- to start the readings of the chart is at
12 o'clock. >> So we wanted people to start reading at
1790, so we put that 12 clock. And people will generally read a radial chart as they
would read a clock. So going from 12 clock to one o'clock to two o'clock and three and
four and around clockwise. >> And that is I think the -- was in a rule
of thumb I think -- we're thinking people that expected people to work really hard if
they were doing anything but starting at 12 clock going clockwise.
>> And I would like to point out that working hard is not what we are after your.
>> It is sort of the opposite of what you're after.
>> That is exactly right. You want -- what we like about the radio graphic is that people
can related to a clock and read it that way. So you absolutely want to make weedy things
as easy as possible. There is sort of a balance between getting people to look at things in
a new way, but still keeping it easy to read and understand me.
>> Going along with that answer, do you run the graphics [ Indiscernible ] is to see if
they can get that at a glance type of understanding from what you have created?
>> Yes -- this is Alex about -- this is Alex. Absolutely. Is important to get your friends
and family involved at the concert stage without -- especially a big usability stage once you
have created something in your testing it. >> We have a team that is putting things together
initially. 18,000 people from international mapping and people at the Census Bureau. And
then it goes around the Census Bureau and people chime in. But those folks at Census
Bureau are pretty much the graphic and data experts. So we went to to people outside of
demography and data that will take a look at it.
>> And we try to do that on every graphic. >> This is Eric again.
>> Also for visualizations that are really interactive tools more than visitation -- there
are so more complex data sets that weird explaining visually at this time. And those tools go
through formal usability testing where we have taken people from the public and down
in front of a machine and asked them to do stuff. And -- we formalize -- when it gets
to a tool that will exist for love, the web. And it will have a lot of viewers with a broad
push knowledge -- with a broad data set -- we make these in a formal usability testing.
>> Great -- that was another question. Thank you for addressing that.
>> Also, you talk little bit about the time it specifically takes to run through that
seven step process? >> Of course. As you might imagine it is quite
variable. And some start with a very good idea and we know how we want to visualize
it. And we are using a widget that we have already created in every thing can be done
in a day. >> I say that is unusual. It usually is a
process of a couple of days. Time spread over the weaker to. We are working on a weekly
schedule. So we have to have many things in the pipeline.
>> We don't always work on them from start to finish. We have several in process.
>> I would say that the range is from half a day to day up to several days.
>> And this is Eric again I would like to add in that Alex takes that long up because
Alex does this for a living rock as we move -- as a -- Alex does this for a living.
>> Your results may vary. As we move to people having people inside the building, we are
giving them a bit more time. >> Okay great.
>> K talk a little bit -- about the consideration that is given to these values of a visualization
-- versus a simple -- and I'm quitting -- quoting here -- for several does anyone ask the question
who cares about the density of population along I-95?
>>[ Indiscernible - multiple speakers ] >> Go-ahead Mark. Now I will just buy another
point here so some of the early visualizations -- you will notice if you look at the gallery,
probably the first two dozen were heavily historical data and almost entirely, if not
entirely from the census. >> And there's a reason for that because we
were essentially leveraging some work that we are even doing. For a different effort.
>> So those visualizations are ones which we had -- which we basically had already drafted
many of them. And if you kind of look at the later ones that we are now branching out into
other data sets we have things that in some cases are not so historically minded. So big
wreck technology and rosacea last week. That assist the most recent estimate. There's nothing
really historical about that one. >> And they definitely very week by week.
In terms of the wow factor. I think there is -- very different reasons and different
values for each one. Some of them you know they really do maybe spark an interest to
get people to them -- to that look at data for other things. Saw them like the one Eric
showed with the eyelids and high income -- you really I mean you can get an awful lot of
that visualization. I don't know whether you would actually need to to consult any of these
tabular data behind it. Because it is essentially all there just in a different form.
>> So it basically buries. But certainly they -- we want there to always be substance behind
them. This not just sort of a wow factor. >> And I would like to chime in. That I don't
understand the question about who would be interested in the population density along
any major roadway. That question does not make sense to me. I work at the Census Bureau.
I care a lot about population density. And they don't understand if you don't.
>> [ laughter ] >> Thank you. We have a couple questions about
the makeup of your team sheet so I am going to just try to lump them together. So here
we go. What advice would you have for someone looking to break into this field -- what tools
or programming linkages would you suggest them mastering. And I would come to these
make up your team? >> Okay -- I will start that one [ Indiscernible
- multiple speakers ] >> I know Eric will have a lot of good answers.
But the first and foremost -- now your data. The first competency is to note your data
and not the analyses and to be able to think about what does data mean.
>> That is -- if it is first, last, middle -- you have got to have that. And a big mistake
that I can -- one of the things we have to fight here is people who are sort of excited
about learning programming or learning the incredibly necessary work of development.
But don't really know -- they have a look at the data. I mean they sort of trust that
if they learn programming, then everything else will flow from it. But it is backwards.
So before we talk about all of the necessary [ Indiscernible ] I want to pitch in that
you have to start with that one >> I think that is a great way to start Eric.
And I would argue that every member of the team -- even if they are not specialists in
the data set or in the subject matter of the data set, needs to take the time to understand
that at least the basics about publishing data. About canonic data. So that they can
better do their role, perform their role in the team.
>> And I think the team would start with a data expert. And this is somebody that is
not just good with numbers, but good with the subject matter of the numbers. That can
be one person or possibly two people. >> You need somebody who is a very skilled
in graphic arts. >> Sort of before the programmer, I would
but the people that can visualize data and graphic ideas into effect using graphic arts
software. >> If you can't make the data visual, that
is the next step after understanding the data -- is making the data jewel.
>> And a simple non-interactive image is often the best way to share the data or a very effective
way. So before you get to the programming, you have to create the graphic images.
>> So you need someone who understands the data. And you need someone to make a visual
-- often a greater -- graphic artist. And you need a programmer or two that understands
the interactivity and can write the code. >> That sometimes this would be to people
-- what would be the interface -- user interface specialist and one that can write the code
that makes things happen. >> Sometimes one person can do both of those
roles. >> So you have got for five people on the
team. And specifically to the question of what software to use -- I sort of laid out
some of the software, but very specifically to a programming language. I would say the
programming thing which to learn is JavaScript. All the interactivity that we have created,
with the exceptions of a couple of flash's limitations, is all done through JavaScript.
>> Anything to add Mark? >> I will echo what Eric started by saying
if you don't understand the data, then you don't know whether it is something -- okay
is it something that you want to map? Is it something that you want to graph? Isn't something
where the story is stored gold -- is it something where the story is historical?
>> The data will tell you all of those things. And if you don't know the range of the values,
it is really hard to kind of get past that. I don't know how you can -- maybe you can
look out. But if you don't have somebody who really knows the basic stories, it is hard
to get past that. Maybe they don't know anything else. Maybe they're not good at the visualization
part or the programming or anything else. But they can at least tell you in no -- these
are the three or four important things to know about this.
>> And we sometimes don't -- we don't like to use the word stories but in fact that it
can but they are a little bit. What are the findings what are the three or four -- if
you are looking at say -- County level income data as you would've the three or four things
that you want to be with to say about the data set?
>> And so if you have the person in the room knows that, everything else is really just
edit the details. >> And it can be hard. Definitely at census
we need more people who come to have a visual skills. But you start the data set knowledge
and everything goes from there. >> Great. Thank you all. I just have one more
question because we are getting close to the time to wrap up. But can you each tell us
outside of your own work I'm a what are some of your favorite pieces of data visualization.
>> Okay well I can start by just talking about various websites I think are worth checking
out and keeping an ion to see what they are doing because they have good visual station
-- data visualization practices. And great post -- practitioners. So certainly Eric mentioned
the New York Times. I would certainly visit their website and a regular basis and check
out the data visualizations that they produce. They have some very good work. Especially
in rats -- and interactive graphs. >> Another newspaper site is the Guardian
and the United Kingdom. They have sort of a data visualization lab. I think is under
their That just as data. >> So those are two really good websites to
check on a regular basis for data digitization. >> Alex -- you stole my site. First and foremost
I was going to say the New York Times. I can't think of any other sort of major was offhand.
I think great data visualizations all over the West -- I mean so much more than even
really two or three years ago it seems to me. Maybe I am more sensitive to it now because
I working closer to it. But it just seems like there is -- there are just lots lots
more great examples out there of things to kind of inspire us. And you might see something
that works really great for one use, and you can just take it and repurpose it for another.
>> Forbes magazine had a great interactive website a couple of years ago. On migration.
That you know you could use that same thing for commuting data or for commodities data
or other stuff. >> And as for me, okay -- I still think the
census bureau atlases from the 1800s were really just tops. But that's because I am
very fond of [ Indiscernible ]. >> There are a few Google the elements -- just
the phrase in the elements -- and you click on images F you are doing an image search.
And you come up with a couple of data digitization. One of which shows the stability of the elements.
And it is a simple -- ages shows stable systems -- protons versus neutrons -- and provides
a color-coded to show how stable the particular element is or the particular formation.
>> And then it shows the one-to-one line. And the one-to-one line -- the scatter plots
diverts from it. So that's the bigger the Adam at the bigger element, the more neutrons
you need to achieve stability. >> I consider this to be the visualization
-- [ Indiscernible ] in the entire universe because it describes the entire universe.
It actually -- the nature of matter. And you can just see -- is neutrons -- that were stability
from an Adam comes from. The whole thing derives from a simple scatter form -- I am completely
astonished by that. So that is my favorite visualization from all-time except possibly
for the Census BureauGovernor that I showed you. Where the people of the United States
come forward to actually form the United States. >> So those are favorites
>> Great. Thank you so much. I would like to thank everyone for attending today. And
if your question was not answered, we would do our best to follow up with the Census Bureau
and get back to you. >> Again, let me thank Eric, Mark and Alex.
Thank you so much for joining us and taking time to present on this topic today.
>> And thank you to Jean Holmes for your brief introduction and reminder on data.gov and
what it has to offer. >> That is the end of our webinar. Thank you
all. >> You're welcome.
>> [ Event concluded ]