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

可變寬度條形圖有不同類型，但最受歡迎的有兩種：1)Mekko條形圖； 2)Marimekko圖表。

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.

它們在以下方面有所不同：雖然Bar Mekko圖表 (BMc)像“標準”條形圖，但是其條形具有可變寬度， Marimekko圖表 (Mc)就像100％堆疊的條形圖，但是其條形是可變寬度。

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.

為何 ：它們用于顯示數據集中每個類別的兩個數字變量； 目標是在類別之間進行比較 ，而不是在數值變量之間進行比較 。 它們廣泛用于儀表板或市場營銷和銷售演示，其中類別通常是產品，區域，部門，細分市場等。數值變量通常是銷售，利潤，成本，利潤，增長率等。它們不適合用于時間分布，關系或趨勢分析。

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 ：是一個類似于標準條形圖的二維圖形，通常以垂直方向排列矩形條。 垂直軸具有數字刻度(在Marimekko圖表上為100％)，代表定量變量之一。 水平軸可以是數字的，也可以是分類的。 如果是數字，則每個矩形的寬度與第二個定量變量的值成比例，并具有不同的顏色和標識它的圖例。 如果是分類的，則每個條的寬度也表示第二個定量變量的值。

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.

對于每個類別，矩形的寬度通常是不同的。 用數字或百分比指示水平軸上表示的數字變量的值始終很方便。 這些值通常在較高的基線上指示。

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.

與標準條形圖不同，條形圖之間不留任何空間 (請注意不要將Mekko圖表與直方圖混淆[https://towardsdatascience.com/histograms-why-how-431a5cfbfcd5])。 在BMc和Mc中， 水平軸的整個寬度都被占用 。

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條形圖是經典條形圖的替代方案，可讓您減少業務演示中的圖數。 為實現此目的，圖表通過條形圖的高度對數值變量之一進行編碼，并通過條形圖的寬度對另一個定量變量進行編碼。 下圖顯示了Mekko條形圖的示意圖：數字垂直軸； 具有可變寬度條的水平軸，每個條代表不同的類別。 產品線，部門或區域是橫軸上指示的類別的示例。 銷售，成本或利潤可以是由矩形寬度表示的數字變量。

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圖表是條形圖的替代方案，可減少在業務演示中的數量：它們顯示與主要類別中的子類別或子類別相關的數字信息。 因此，這種類型的可視化使我們可以為每個類別和每個子類別顯示兩個數值。

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.

通過將其分成較小的矩形，它們占據了整個矩形區域。 每個類別分為由堆疊矩形表示的子類別。 這些矩形中的每個矩形都對數據集的每個子類別的數值進行編碼。 它們在視覺上等同于堆疊的條形圖，在該條形圖中，除了具有不同高度的矩形段之外，我們現在還具有可變的條形寬度。

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.

圖的垂直軸顯示百分比比例，水平軸被標準化以占據圖的整個寬度，而在條之間不留任何空白。

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.

圖2在一張圖表中描述了一家公司的年收入，這些公司分為不同的品牌，以及對應于不同地區的百分比。 可以看到表征Marimekko圖表的元素：將矩形區域劃分為寬度不同的較小矩形； 垂直堆疊的矩形； 橫軸占據圖表的整個寬度； 帶有百分比刻度的垂直軸； 在最高基準上每個品牌的總收入； 不同的條形寬度，可以計算每個品牌對總收入的相對貢獻。

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).

詞源：Marimekko是芬蘭的一個紡織品牌，具有多種多樣的形狀圖案，以及簡單明亮的顏色和樣式(＃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圖表和Marimekko圖表都難以閱讀和解釋，因為它們基于觀眾通過比較區域對數字信息進行解碼的能力。 必須記住，人類擅長評估距離而擅長計算面積。 此外，隨著矩形數量的增加，我們對不同區域進行適當比較的能力也會降低。

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圖具有與堆積條形圖相同的缺點：a)子類別的數量存在實際限制，超過該限制，可視化將變得困難； b)在上下基線附近的子類別之間進行最佳比較。 隨著我們遠離它們，很難評估子類別之間的差異。

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

重復使用Mekko圖表時要小心：記住矩形條非常“重”并且是主要的視覺標記。

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.

請注意，觀眾通常對傳統的條形圖非常熟悉，但對可變寬度的條形圖則不太熟悉。 如果您的演示文稿中包含Mcs或BMcs，請花一些時間向聽眾解釋它們的特殊特征，以免影響故事的講述。

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

Marimekko圖表不能顯示負值，也不能將絕對值和相對值組合在一起。

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.

一些可視化工具允許您在垂直和水平布局之間切換。 然而，在通常的實踐中，由于水平布置不必要地使講故事復雜化，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.

不要將Mcs與spineplots混淆 。 在其最嚴格的定義中， 旋轉旋轉圖是一維的水平堆疊的條形圖，用于在列聯表(＃3)中顯示兩個交叉分類的分類變量的頻率，比例或百分比。 之所以會出現這種混亂，是因為某些旋轉圖的可視化工具允許垂直定向，他們將其稱為鑲嵌圖 。 該術語也被誤認為是Marimekko圖表，但應保留給那些允許使用可變寬度矩形檢查兩個或多個分類變量之間關系的圖表。

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中找不到允許直接繪制Mcs或BMcs的繪圖函數。 Stack Overflow用戶編碼了一種策略，該策略使用條形圖通過Matplotlib繪制Bar Mekko圖，使用寬度參數(＃4)分別修改每個條的寬度 。 我對其代碼進行了少量修改，以得到下圖，非常接近Bar Mekko圖表。

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允許從列聯表( statsmodels.graphics.mosaicplot.mosaic )創建鑲嵌圖形。 下圖和繪制它的代碼可以在statsmodels組織的以下頁面中找到 ：

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):

如下圖所示，鑲嵌圖在水平和垂直軸(＃5)上均指示類別變量：

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.

總結：在自己適當地加權數值變量之一的大小，能力有關美康圖表謊言最重要的概念想法。 傳統條形圖的所有變體(標準，分組，堆疊，重疊等)對所有類別都使用相同的條形寬度，因此只能通過單個數字變量的長度或高度進行比較。 Mekko圖表中的變量寬度使我們可以在單個圖形中顯示第二個數字變量的相對值。 相反，它們的視覺效果很差 ：到達受眾的編碼信息的準確性和清晰度相對較低。

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

如果您發現這篇文章感興趣，請閱讀我以前的文章：

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

“直方圖，原因和方式，講故事，提示和擴展”

“All You Need to Know About Bar Graphs”

“您需要了解的有關條形圖的所有信息”

#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：北卡羅來納州福克斯(Fox，N.)，“講故事的國家：Spineplots及其親屬”，《國家統計》(The Stata Journal)(2008)，第8期，第1期，第105–121頁

#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