There are different types of variable width bar charts but two are the most popular: 1) Bar Mekko chart; 2) Marimekko chart.

可变宽度条形图有不同类型，但最受欢迎的有两种：1)Mekko条形图； 2)Marimekko图表。

They differ in the following: while the Bar Mekko chart (BMc) is like a “standard” bar chart, but its bars are of variable width, a Marimekko chart (Mc) is like a 100% stacked bar chart but its bars are of variable width.

它们在以下方面有所不同：虽然Bar Mekko图表 (BMc)像“标准”条形图，但是其条形具有可变宽度， Marimekko图表 (Mc)就像100％堆叠的条形图，但是其条形是可变宽度。

WHY: they are used to show two numerical variables for each category present in the data set; the goal is to make a comparison between the categories, but not between the numerical variables. They are widely used in Dashboards or in Marketing and Sales Presentations, where the categories are usually products, regions, sectors, segments, etc. The numerical variables are typically sales, profits, costs, margins, growth rates, etc. They are not suitable for Distribution, Relationships, or Trends over Time analysis.

为何 ：它们用于显示数据集中每个类别的两个数字变量； 目标是在类别之间进行比较 ，而不是在数值变量之间进行比较 。 它们广泛用于仪表板或市场营销和销售演示，其中类别通常是产品，区域，部门，细分市场等。数值变量通常是销售，利润，成本，利润，增长率等。它们不适合用于时间分布，关系或趋势分析。

HOW: is a two-dimensional graph similar to a standard bar chart, with rectangular bars usually arranged in a vertical orientation. The vertical axis has a numerical scale (100% on a Marimekko Chart) and represents one of the quantitative variables. The horizontal axis can be either numerical or categorical. If it is numerical, each rectangle has a width proportional to the value of the second quantitative variable, a different color, and a legend that identifies it. If it is categorical, the width of each bar also indicates the value of the second quantitative variable.

HOW ：是一个类似于标准条形图的二维图形，通常以垂直方向排列矩形条。 垂直轴具有数字刻度(在Marimekko图表上为100％)，代表定量变量之一。 水平轴可以是数字的，也可以是分类的。 如果是数字，则每个矩形的宽度与第二个定量变量的值成比例，并具有不同的颜色和标识它的图例。 如果是分类的，则每个条的宽度也表示第二个定量变量的值。

The width of the rectangles is generally different for each category. It is always convenient to indicate with numbers or percentages the value of the numerical variable represented on the horizontal axis. These values are usually indicated on the upper baseline.

对于每个类别，矩形的宽度通常是不同的。 用数字或百分比指示水平轴上表示的数字变量的值始终很方便。 这些值通常在较高的基线上指示。

Unlike the standard bar chart, no space is left between the bars (Be careful not to confuse Mekko charts with Histograms [https://towardsdatascience.com/histograms-why-how-431a5cfbfcd5]). In both, BMc and Mc, the entire width of the horizontal axis is occupied.

与标准条形图不同，条形图之间不留任何空间 (请注意不要将Mekko图表与直方图混淆[https://towardsdatascience.com/histograms-why-how-431a5cfbfcd5])。 在BMc和Mc中， 水平轴的整个宽度都被占用 。

The Mekko Bar chart is an alternative to classic bar charts allowing you to reduce the number of charts in a business presentation. To achieve this, the chart encodes one of the numerical variables by the height of the bars and the other quantitative variable by the width of the bars. The following graph shows a schematic representation of a Mekko Bar chart: a numerical vertical axis; a horizontal axis with variable width bars, each representing a different category. Product lines, sectors, or regions are examples of categories indicated on the horizontal axis. Sales, costs, or profits could be numerical variables represented by the width of the rectangles.

Mekko条形图是经典条形图的替代方案，可让您减少业务演示中的图数。 为实现此目的，图表通过条形图的高度对数值变量之一进行编码，并通过条形图的宽度对另一个定量变量进行编码。 下图显示了Mekko条形图的示意图：数字垂直轴； 具有可变宽度条的水平轴，每个条代表不同的类别。 产品线，部门或区域是横轴上指示的类别的示例。 销售，成本或利润可以是由矩形宽度表示的数字变量。

The Marimekko Chart is an alternative to grouped bar charts allowing to reduce their number in a business presentation: they show numerical information related to subgroups or subcategories within the main categories. Therefore, this type of visualization allows us to show two numerical values for each category and each subcategory.

Marimekko图表是条形图的替代方案，可减少在业务演示中的数量：它们显示与主要类别中的子类别或子类别相关的数字信息。 因此，这种类型的可视化使我们可以为每个类别和每个子类别显示两个数值。

They occupy an entire rectangular area by dividing it into smaller rectangles. Each category is divided into subcategories indicated by stacked rectangles. Each of these rectangles encodes the numerical values for each subcategory of the data set. They are visually equivalent to stacked bar charts, where in addition to having rectangular segments of varying heights, we now also have variable bar widths.

通过将其分成较小的矩形，它们占据了整个矩形区域。 每个类别分为由堆叠矩形表示的子类别。 这些矩形中的每个矩形都对数据集的每个子类别的数值进行编码。 它们在视觉上等同于堆叠的条形图，在该条形图中，除了具有不同高度的矩形段之外，我们现在还具有可变的条形宽度。

The vertical axis of the graph displays a percentage scale while the horizontal axis is normalized to occupy the entire width of the diagram without leaving any blank space between the bars.

图的垂直轴显示百分比比例，水平轴被标准化以占据图的整个宽度，而在条之间不留任何空白。

Figure 2 describes in a single graph the annual income of a company separated into its different brands and the percentages corresponding to the different regions. The elements that characterize a Marimekko chart can be seen: a rectangular area divided into smaller rectangles of varying width; vertically stacked rectangles; a horizontal axis that occupies the entire width of the chart; a vertical axis with a percentage scale; total revenues per brand on the top baseline; different bar widths that allow calculating the relative contributions of each brand to the total revenues.

图2在一张图表中描述了一家公司的年收入，这些公司分为不同的品牌，以及对应于不同地区的百分比。 可以看到表征Marimekko图表的元素：将矩形区域划分为宽度不同的较小矩形； 垂直堆叠的矩形； 横轴占据图表的整个宽度； 带有百分比刻度的垂直轴； 在最高基准上每个品牌的总收入； 不同的条形宽度，可以计算每个品牌对总收入的相对贡献。

Source of the word: Marimekko is a Finnish textile brand that has large and varied shape patterns, as well as simple and bright colors and styles (#2).

词源：Marimekko是芬兰的一个纺织品牌，具有多种多样的形状图案，以及简单明亮的颜色和样式(＃2)。

Warnings:

警告：

Both Bar Mekko Charts and Marimekko Charts are difficult to read and interpret because they are based on audiences’ ability to decode numerical information by comparing areas. It is essential to keep in mind that humans are good at evaluating distances and bad at calculating areas. In addition, as the number of rectangles increases, our ability to make a proper comparison of the different areas decreases.

Bar Mekko图表和Marimekko图表都难以阅读和解释，因为它们基于观众通过比较区域对数字信息进行解码的能力。 必须记住，人类擅长评估距离而擅长计算面积。 此外，随着矩形数量的增加，我们对不同区域进行适当比较的能力也会降低。

Marimekko diagrams suffer from the same drawbacks as stacked bar charts: a) there is a practical limit to the number of subcategories, beyond which visualization becomes difficult; b) the best comparisons are made between subcategories near the lower and upper baselines. As we move away from them, the differences between the subcategories become very difficult to evaluate.

Marimekko图具有与堆积条形图相同的缺点：a)子类别的数量存在实际限制，超过该限制，可视化将变得困难； b)在上下基线附近的子类别之间进行最佳比较。 随着我们远离它们，很难评估子类别之间的差异。

Be cautious with the repeated use of Mekko charts: remember that rectangular bars are very “heavy” and dominant visual markers.

重复使用Mekko图表时要小心：记住矩形条非常“重”并且是主要的视觉标记。

Note that audiences are usually very familiar with traditional bar charts but not with variable width ones. If you have Mcs or BMcs in your presentation, take some time to explain to the audience the particular characteristics of them so as not to impair the storytelling.

请注意，观众通常对传统的条形图非常熟悉，但对可变宽度的条形图则不太熟悉。 如果您的演示文稿中包含Mcs或BMcs，请花一些时间向听众解释它们的特殊特征，以免影响故事的讲述。

Marimekko Charts cannot show negative values, nor can they combine absolute values with relative values.

Marimekko图表不能显示负值，也不能将绝对值和相对值组合在一起。

Some visualization tools allow you to switch between a vertical and a horizontal layout. However, in the usual practice the vertical layout is used since the horizontal one complicates the storytelling unnecessarily.

一些可视化工具允许您在垂直和水平布局之间切换。 然而，在通常的实践中，由于水平布置不必要地使讲故事复杂化，因此使用垂直布置。

Do not confuse Mcs with spineplots. In its most rigorous definition, a spineplot is a one-dimensional, horizontal stacked bar graph used to display frequencies, proportions, or percentages of two cross-classified categorical variables in contingency tables (#3). The confusion arises because some spineplots visualization tools allow a vertical orientation, which they call mosaic plots. This term is also erroneously attributed to the Marimekko Charts but should be reserved for those charts that allow examining the relationship between two or more categorical variables by means of variable-width rectangles.

不要将Mcs与spineplots混淆 。 在其最严格的定义中， 旋转旋转图是一维的水平堆叠的条形图，用于在列联表(＃3)中显示两个交叉分类的分类变量的频率，比例或百分比。 之所以会出现这种混乱，是因为某些旋转图的可视化工具允许垂直定向，他们将其称为镶嵌图 。 该术语也被误认为是Marimekko图表，但应保留给那些允许使用可变宽度矩形检查两个或多个分类变量之间关系的图表。

I did not find plotting functions in Python that allow direct graphing of Mcs or BMcs. A Stack Overflow user coded a strategy to plot Bar Mekko charts with Matplotlib using bar charts to which the width of each bar is individually modified using the width parameter (#4). I made minor modifications to his code to get the following figure, a fairly close approximation to a Bar Mekko chart.

我在Python中找不到允许直接绘制Mcs或BMcs的绘图函数。 Stack Overflow用户编码了一种策略，该策略使用条形图通过Matplotlib绘制Bar Mekko图，使用宽度参数(＃4)分别修改每个条的宽度 。 我对其代码进行了少量修改，以得到下图，非常接近Bar Mekko图表。

On the other hand, version 12.0 of statsmodels allows the creation of mosaic graphics from contingency tables (statsmodels.graphics.mosaicplot.mosaic). The following figure and the code to draw it can be found in the following page of the statsmodels organization:

另一方面，版本12.0的statsmodels允许从列联表( statsmodels.graphics.mosaicplot.mosaic )创建镶嵌图形。 下图和绘制它的代码可以在statsmodels组织的以下页面中找到 ：

https://www.statsmodels.org/stable/generated/statsmodels.graphics.mosaicplot.mosaic.html

https://www.statsmodels.org/stable/generation/statsmodels.graphics.mosaicplot.mosaic.html

As can be seen in the following figure, mosaic graphs indicate categorical variables on both the horizontal and vertical axis (#5):

如下图所示，镶嵌图在水平和垂直轴(＃5)上均指示类别变量：

To summarize: the most important conceptual idea related to the Mekko Charts lies in their ability to appropriately weight the magnitude of one of the numerical variables. Traditional bar charts in all their variants (standard, grouped, stacked, overlapping, etc.) use the same bar width for all categories and therefore only make comparisons by means of the length or height of a single numerical variable. The variable width in the Mekko Charts allows us to show in a single graph the relative value of the second numerical variable. In contrast, they have low visual effectiveness: the accuracy and clarity of the coded information that reaches the audience is relatively low.

总结：在自己适当地加权数值变量之一的大小，能力有关美康图表谎言最重要的概念想法。 传统条形图的所有变体(标准，分组，堆叠，重叠等)对所有类别都使用相同的条形宽度，因此只能通过单个数字变量的长度或高度进行比较。 Mekko图表中的变量宽度使我们可以在单个图形中显示第二个数字变量的相对值。 相反，它们的视觉效果很差 ：到达受众的编码信息的准确性和清晰度相对较低。

If you find this article of interest, please read my previous:

如果您发现这篇文章感兴趣，请阅读我以前的文章：

“Histograms, Why & How, Storytelling, Tips & Extensions”

“直方图，原因和方式，讲故事，提示和扩展”

“All You Need to Know About Bar Graphs”

“您需要了解的有关条形图的所有信息”

#1: https://vizzlo.com/

＃1： https ： //vizzlo.com/

#2: https://en.wikipedia.org/wiki/Marimekko

＃2： https ： //en.wikipedia.org/wiki/Marimekko

#3: Fox, N., “Speaking Stata: Spineplots and their kin”, The Stata Journal (2008), 8, Number 1, pp. 105–121

＃3：北卡罗来纳州福克斯(Fox，N.)，“讲故事的国家：Spineplots及其亲属”，《国家统计》(The Stata Journal)(2008)，第8期，第1期，第105–121页

#4:https://stackoverflow.com/questions/57164333/bar-chart-with-different-widths-and-colors-in-matplotlib-bar-mekko-chart

＃4：https：//stackoverflow.com/questions/57164333/bar-chart-with-different-widths-and-colors-in-matplotlib-bar-mekko-chart

#5: https://www.statsmodels.org/stable/generated/statsmodels.graphics.mosaicplot.mosaic.html

＃5： https ： //www.statsmodels.org/stable/generation/statsmodels.graphics.mosaicplot.mosaic.html