dylanjm7:45:002Interesting to me that hue doesn't have a natural ordering. It's highly context specific.#ERROR!
hathawayj12:10:0011Humans see the size of something according to its area not its diameter or radius.4
mcgurkk7:33:001It is OK to use two charts. The Gestalt principles were quite interesting and great to keep in mind. People like a picture they can see. "Above all else, show the VARIATION IN data"23
mafubelut7:15:001Erase non data ink that interferes with detection or doesn’t assist assembly and estimation. Compared to what? Most important question in quantitative analysis. 'Pie-charts are a joke'
carmonas11:39:0011The eye can give a quantitative estimate of saturation. Standardizing data lets us compare anomalies between series. Good plots use continuity.10
craigt0:19:3810.33Pi charts are terrible. You should never use them6
seguraj0:07:300.831The explanation of the difficulty to understand colors. I find this interesting as I am color blind.9
spearc16:41:002.53Stacking is almost never the right option. Position on a common scale, even with multiple charts, is preferable.3
spearc35:05:0013Make sure grid lines don't interfere with detection.2
spearc25:26:001.53Growth charts should directly plot growth (slope).10
gibbardt11:11:0013Pie charts should never be used. color, hue and saturation should be used carefully, bar charts should always show 0 to be able to compare lenth.7
foglea6:40:000.333Rankings of encoding2
foglea5:27:0013Three levels of estimation should be used when plotting data22
foglea26:05:00153Plotting growth directly makes a noticeable difference27
millerc7:50:004:303Coloring can be a great to present your graph without making multiple graphs11
orrc25:30:001.51Growth charts should display slope not points. "If you care about growth plot it directly."10
orrc6:30:0011The list displayed is the order with which you should try to present your most important data first to last. (Using multiple charts to hit these points with important data is ok.)15
millerc21:34:0011comparison is trivial on a common scale5
orrc29:10:0011The law of continuity is a naturally helpful and useful even vaguely suggestive continuity is indicative to the human mind of relationship.6
millerc24:47:0011talking about why scatter plots are so beneficial in showing relationships between 2 metrics2
grimauds11:05:0011Saturation is easier to interpret data with than hue2
grimauds38:12:0011If you are using angle to compare angle, try to make the average angle 45 degrees, humans can best interpret angle at 45 degress8
grimauds22:38:0011You can standardize time series data to make it easier to compare6
romerob4:020.51We need to understand what operation we are trying to convey to the audience; understand the purpose of each of these three operations19
omokohm5:300.91At the heart of quantitative reasoning is a single question: Compared to what? I feel this is profound in visualizing data to an audience. If the audience can't understand this from your data(visualization), one has failed in converying a story to your audeince.10
romerob35:17:000.51The difference is only as impactful as the proportion conveys -- small percent changes are less noticeable, even if the absolute value difference is greater24
romerob25:27:0011Different charts highlight different things -- showing the literal population accomplishes an entirely different purpose than showing the relative growth13
fullmert1:40:0010:36He highlights the point that our eyes transmit information directly to our brain. This makes it much easier to interpret rather than numbers.27
fullmert16:15:0010:36The point of visualization is to make ranking and discrimination trivial for anyone looking at the data.18
gilesc28;00:0030:36The law of continuity states that you organize your data based on data points rather than the names of data, that way the data is organized the same way your brain thinks.
gilesc6:27:001:450:36What Clevend says is the most important aspects of vizulizations/ graphs, scale needs to be accuratley presented.
gilesc20:17:0020:36Instead of using pie charts tables should be used instead, especially when using small amounts of data.
ferrinb6:20:0010:36This table is a great reference to summarize this presentation
ferrinb18:07:0010:36It's better to use two charts than to try to make a chart that has too much information. Even if you don't have room for two charts, it's better to only use one chart that says some information well than one chart with lots of information that's hard to understand
ferrinb25:30:0020:36If you want to compare the slopes of things, plot the slopes (the derivatives) on a common scale
Lianga6:41:0010:36how to make readable format and what's effective
lianga16:58:0020:36different ways of displaying counts
lianga20:07:0020:36tables are better than graphic for small data
ackleyn2:00:0021The idea of pre-attentive processing is pretty interesting. You can display data in multiple ways, and get completely different results. Some being better than others.
ackleyn27:08:001.51Gestalt Psychology is an interesting technique we can apply to data science/analytics
ackleyn31:58:0011Grouping certain things together can make it easier for the viewer to understand visuals
pohlsanderc1:23:0021Graphs are much better than tables in most circumstances, because our minds can easily process information in the medium of a chart or graph, while tables are not good for visualizing.
pohlsanderc20:15:0021In small data sets, a table is usually the best option for visualization. Pie charts are garbage in every situation.
pohlsanderc34:59:000.51When visualizing data, there should be a good amount of contrast between the data visualized and the background.
rosen24m50s1mhathawayjA good example how better data presentation could help in the real world in an expanding service
rosen31m30s10shathawayjGood advice to add redundancy to channels through which we display data
rosen35m19s1m30shathawayjShowing how porportional differences are more important than absolute differences
Dobbss20:17:002hathawayjThe explanation of why Pie Charts are never a good idea really stuck with me. I have often used them and definitely see his reasoning behind not using them
Dobbss25:00:003hathawayjHere he explains why he usually graphs growth. He explains that if growth is what you are showing plot it directly. I was impressed with how I couldnt see the growth well before he directly plotted it.
Dobbss7:33:001hathawayjHere he explains that more often than not using 2 plots works better than overlaying multiple metrics on one chart. He explains why stacked bar charts are often so hard to read and understand.
garcias6:40:000.84hathawayjWhen make a graphic we should think? What is the single most important comparision we want a user to make
garcias17:30:001.27Instead of Using one single graphic combine two visuals together, it gives more clarity
garcias29:13:000.33"The law of Continuity" First image was difficult to analize the second was easy to group becasue of the established directions
porterm19:34:002Pie charts are always the wrong choice, they look highly unprofessional and difficult for a brain the easily comprehend. Use tables instead for small datasets.
porterm25m26s1Growth charts are often misrepresenting and showing little to no useful information, growth chart is often best when looking at percentages.
35m2Webers law is why grid lines can be useful on a plot, due to aiding the user into seeing porportions.
mendenhallz22:10:001When two things need to go on the same graph but their scale is way different, use a standardized value. Subtract it's mean and divide by its standard deviation.
mendenhallz25:27:001.5Growth charts usually aren't. If the slope is more important than plot that, not the actual growth. It will help you see the actual growth better.
mendenhallz29:15:000.5Law of continuity is key. You want continuous patterns because that is what will make a chart easy to read. Things that are scattered all over are useless.
monterrosod6:57~ 8 secondsWhen making a graphic we have to think about what is the most important comparison we want our audience to make.
monterrosod18:34:000.33instead of trying to make just one graphic that shows everything (which it might not show what we want well) we should make multiple charts
monterrosod19:352pie charts use angles to show data which is forth on the list. Therefore, there is always a better way to show the data.
deloayzag11:12:001Saturation has two advantages, first it has a natural order, second the eye can give a quantitative estimate of saturation. Guesses can be made and get at least one significant figure out of the guess.
deloayzag16:451Stacking charts make information hard to read. Instead use more charts to explain the different sets of data. If needed decide which is the more important comparison and leave out the other views.
deloayzag21:381.3Non-align scales can be standardized by subtracting its mean and dividing by the standard deviation. The absolute information however is lost so you need both views available.
dingI11:07:001The two advantages of the color saturation taught me there is no perfect measurement for data analyzing, each measurement has its own advantages in some special field.
dingI243We should select the graph depending on what information we want to see or we want to compare.
dingI37:50:002Banking to 45, the average slop is 45. It is easy to see the variation.
aldert16:442I like how he says that anything stacked is almost always a mistake. He then goes on to use two graphs side by side and it made more sense then the diamonds stacked on each other did.
aldert19:351Pie charts do not serve a purpose for telling a story of data. They easily distort data and can be very difficult to read. He also mentions that all good pie charts are jokes. He then shares the pyramid pie chart.
aldert30:11:001At this moment, he is talking about the usefulness of hue and he is also talking about the law of similarity. The different hues make our brains collect the data points into one group.
rodrigueze5:30:002All the levels of estimation involves comparison, this brings more significance a power to any study.
rodrigueze7:00:001everytime we make a graph, we have to ask ourselves what is the most important comparison that I want the user to make? by undestanding human color perception we can make our graphs and plots more powerful and meanningful to our audience.
rodrigueze281reification is when we experience more information that is presented on the big picture.
stuarte16:42:00"Stacked anything is nearly always a mistake"
stuarte19:34"Pie charts are always a mistake" I will admit, I didn't realize how hated pie charts were, professionally. I thought that the pie chart he showed with the pyramid and sky was funny.
evansl1:482Pre-Attentive Processing - best visualizations utilize pre-attentive processing. I love this point b/c it emphasizes just how simple visualizations should be in order to be grasped by th audience. More complicated does not always equal more impressive.
evansl8:00:003Color coding by hue is often a bad idea as it is often impossible to tell the differences between hues AND hues have no order. Saturation is much better than hue as it has a natural order
evansl20:002Pie charts are never preferable because the human eye/mind is terrible and judging differences between angles and slopes. This applies to any chart. There is often a better way to present data than with an angle.
stuarte8:19:00I like the way that he shows here how difficult a chart can be to read if not organized properly.
catherb0:34:005 secondsThe goal of visulization is to make new ideas appear in the minds of others.
catherb13:5845 secondsIt's interesting how the angle of a graph is in the mind of people is already being perceived as having a slope is a good thing.
catherb18:59:0030 secstacked charts are almost always a mistkae, as they don't show the postion only a lenght for comparison.
shugaa4:3512three levels of estimation "When you're trying to think with data, you are always making comparisons" / ranking of types of comparisons in charts
shugaa16:452stacked bars vs. parallel coordinates plot (also, sometimes it's better to just use two plots instead of cramming everything into one)
shugaa23:110.5sometimes one type of plot is better, and sometimes another plot is better (i.e. two separate plots vs. one plot on a common scale) depending on what the purpose is and what you want the viewer to understand from looking at it.
hansenc6:1610The difference between how humans make estimations on data, and that some forms are better than others.
hansenc29:134Sorting or Categorizing the data will help the viewer see trends, and truths in the data.
hansenc35:313When comparing items it easier to compare differences when the differences are a higher percetage of the observation. Webers Law
smithr18:230.5use2 charts if you need 2, if you don't have room, edit. It's hard but it's what viewers need you to do.
smithr4:450.5efficient comparison between data points is almost always the reason for visualization.
smithr26.120.5If growth is important, plot it directly.
villalvap0:240.5Visualization is communication. Our goal is to make new ideas appear in the heads of other people. Good and true statement.
villalvap26:517The Assembly principle. The grouping of the graphical elements.
villalvap19:352Do not use pie charts, They are not useful.
niyonkurue0:55:001:30Visualization is about turnig raw data into images/graphics to communication a certain truth.
niyonkurue3:40:001:25A graph is simply an encoding of data.