Machine Learning

Dangers of decisive charts confusing data and misleading articles

“You do not have to be a professional to deceive someone, even though you can need some technology to see it honest when you are deceived.”

When my coach and I am starting our quarter study with the deceased papers in the industry for data training in the University of Washington, emphasizes the above point from our students. With the early technological priority, improving good and convincing claims with data is easier than ever. Anyone can do something that seems to pass you, but contains the oversight that allows wrong and dangerous. In addition, there are malicious characters who work diligently want Deception, and who have read some of the best ways to do.

I usually start this expression with a little quip, I look so much about my students and ask two questions:

  1. “Is it a good thing if someone dumps you?”
  2. After the general sighing of the confusion followed by a contract that the field play is really bad, asking the second question: “What is the best way to make sure no one else has ever seen it?”

Students usually reflect on that second question for a while, before cleaning a little and seeing the answer: You have to learn how people enlighten you in the first place. Not so then you can use others, but to protect others from making good use.

The same applies to the error zone and a denform. People who want to mislead the details are empowered with many tools, from the highest internet to social media so that they are recently, lying models. To protect yourself from misleading, you need to read their tactics.

In this article, I take important ideas from my Data Visionization Unit with deceptive drawn by the best book Alberto Cairo How many areings of-Has encouraged them to be certain principles about corruption and data. My hope is that you read it, read it, and take it with you analyzing yourself with false attacks for people with bad goals.

People can't interpret the area

At least, not and we translate other visual indicators. Let us simply illustrate this by example. Say we have a very simple set of number number; It is one size and contains just two amounts: 50 and 100. Another way to show this in view of bolds, as follows:

This is true of basic data. The length is the amount that has a great size, and doubles to show double the value. But what happens if we want to represent the same data by circles? Yes, circles are not defined in length or wide. One option to resched up Radius:

Hmm. The first circle has a 100 pixels, and the second has the 50 pixels – so this is ok with technology if we duplicate a Rediard. However, because of the way the area is calculated (πRR²), repeatedly, repeatedly. So what if we try to do that, because it seems more accurate? Here is a revised version:

We now have a different problem. A larger circle is double in a small place, but there is no more you look That way. In other words, though it is a very visible comparison of the amount of double value, people's eyes have difficulty in identifying.

The debate here is trying to use the area as a time marker for the first time. Is not really badBut it is confusing. We increase the number of four-sided quantities, but the area is a two-size number. In a person's test, it will always be difficult to translate accuracy, especially when compared with a natural native representation such as barriers.

Now, this may seem like it is not a great thing – but let's look at what happened when you extend this to the real set of data. Below, I attach two charts to Altair (Python-based view package). Each chart indicates the highest temperature of the first week of 2012 in Seattle, in the USA. The first uses the bar length to make comparisons, and for the second time using round areas.

What makes it easier to see the difference? Myth Held of Second, but if we are honest, it is a lost cause. It is very easy to make straight comparisons and bars, even in the area where we have such data restricted.

Remember that the point of viewing is to clarify data – to make it easier for you to see a common person. To achieve this goal, it is best to use visual logs that make easy process.

Note Political Articles (In any way)

There is a little solemn question I sometimes ask my students in the school assignment around fourth week of the classroom. Assignment is very involved in producing visual epython – but for the last question, I give them a chart where I personally was produced by this one question:

Question: There is one wrong thing with a shading above, an error that does not forgive data recognizing. What's up?

Most think that it has something related to axes, marks, or another feature, often suggests progress as filling circles or making the most educated Aksik labels. Those are good suggestions, but not very stressful.

The most faulty defect (or lack of it, rather) on the above chart is Lost title. The topic is important for a successful witness. Without it, how should we know what this sense is? Currently, we can only know that there should be something to do with something at the levels of carbon dioxide across the age. That's not much.

Many people hear that this requirement is very strong, including eye-sensitive, as part of a large article or other issue or anything associated with the text. Unfortunately, this line of thinking is very straightforward; In fact, visualization must stand alone, because it is usually the only thing that people view – and in social media, the only sharp sharp. As a result, the topic of self-define.

Of course, the title of this clause tells you to be aware of such articles. That is true. While they are needed, they are both sides. Since the architects simply recognize observers know that viewers will pay attention to the title, what emotions say can use it and spend people in limited indicators. Let's look at example:

The above is a photo stolen by the White House Public Twitter in 2017. This picture is also directed by Alberto Cairo in his book, emphasizing many points to do.

First things first. The word “tables of tables,” is referred to in the form of a family based migratory immigration (where the immigration can support many of the United States), criticized by many men and makes the legal opponents heard threatening without threat.

Of course, politics are their natural divine divisive nature, and possible in any side to make a burning argument. The main issue is actually related to one-specific data, that the use of “Chain” means in the context of the shared chart. “The chain” appears to indicate that people do not grow old in a row, in a strange stream and unanswered family relationship. The truth is, of course, that one person can just support family members, and even that takes less time. But when a person reads the phrase “series” and immediately looks at a virtual chart showing it shows, it is easy for a person to suddenly come from an impoverished level.

That Has the problem is any kind of political topic – it makes it very easy to hide in hiding, inaccurate by considering real data, analysis, and eye review.

Here no Data below the above chart. There is no. Zero. Completely organized, and that is not good in a commercial chart in the purpose of appearing to show something meaningful and a lot.

Like a small rabbit rabbit decline in the political debit, here is link to the bottom, the Twitter account that sends the most limited drawings shown by the Congress Floort conference.

Do not use 3D. I'm up.

I will finish this article with a little simple article – but I am important. Under conditions that are not under – no one – to work on 3D chart. And if you're in the spectator's shoes – that is, when you look at 3D pie chart made from someone else – trust you.

The reason for this is simple, and it connects back to what I have discussed by gatherings and rectangles: the size of the third too Really distorts after which they are often the means of great size. Location is already difficult to translate – How much more do you think of a person's eye in volume?

Here is a 3D pie chart I'm produced in random numbers:

Now, here is the same pie chart, but in two dimensions:

Note that the blue is exactly like a 3D version seems to lift, and that red and orange are close to each other in size than before. I also deliberately eliminates leakeels. If you read this article about analytical eye, you probably think that he doesn't do what a big difference. But the fact is, you often see such charts in matters or social media, and speed view everything will ever find.

It is important to ensure that the account has been reported to the fast-tracking.

The last thoughts

Data science is usually taken as full statistical statistics, Computing, and community, how to find and share deep and meaningful vision for a heavy world. This is true – but as the ability to share such an understanding, such extends, so should be our strength to interpret it accurately. It is my hope that in the light, you got this primer to be useful.

Stay watching part 2, where I will speak for a few deceptive strategies and participate in nature – including a mathematical rating, (UN) reliever mathematics, and related steps.

In the meantime, try not to be deceived.

Source link

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button