Stata绘图：柱状图专题

🍎 连享会主页：lianxh.cn

🎦 2021 效率分析专题

📅 2021 年 5.15-16；5.21-22

🔑 主讲：龚斌磊(浙江大学)；连玉君(中山大学)；张宁(山东大学)

🍓 课程主页：https://gitee.com/lianxh/TE

📗 温馨提示：可以分时段报名

New！ lianxh 命令发布了：   GIF 动图介绍

随时搜索 Stata 推文、教程、手册、论坛，安装命令如下：

. ssc install lianxh

作者： 徐阳 (武汉大学)

E-Mail: [e-mail]

Source: 世界银行 stata-visual-library。

注：本推文相关数据可通过上述世界银行网址获取，也可以通过如下两个途径获取：

目录

0. 引言

1. 多变量水平柱状图（Horizontal bar with multiple variables）

2. 两变量柱状图（Bar plot of two variables）

3. 两轴组合柱状图（Combined bar plots with two axes）

4. 并排水平柱状图（Side by side horizontal bar plot）（基于命令betterbar）

5. 两变量堆积柱状图（Stack bar graph by two variables）

6. 并排水平柱状图（Side by side horizontal bar plot）（基于命令weightab）

7. 带有变量分组、标准误差条和跨组比较的水平柱状图（Horizontal bar plot with grouping of variables, standard error bars, and cross-group comparisons）

8. 横向堆积柱状图（Horizontal stack bar plot）

9. 相关推文

温馨提示： 文中链接在微信中无法生效。请点击底部「阅读原文」。或直接长按/扫描如下二维码，直达原文：

0. 引言

本文参考世界银行 stata-visual-library 提供的数据及代码，通过八个范例来介绍如何 绘制柱状图。部分范例有较长的数据生成过程，本文重点在于介绍如何绘制柱状图，故未做详细介绍。

1. 多变量水平柱状图

代码如下：

global graph_opts ///

title(, justification(left) ///

color(black) span position(11)) ///

graphregion(color(white)) ///

ylabel(,angle(0) nogrid) ///

xtitle(,placement(left) justification(left)) ///

yscale(noline) xscale(noline) ///

legend(region(lcolor(none) fcolor(none)))

global pct `' 0 '0%' .25 '25%' .5 '50%' .75 '75%' 1 '100%' ''

quietly do 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/ado/betterbar.ado'

use 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/data/bar-better.dta' , clear

betterbar ///

??_correct checklist refer med_any med_class_any_6 med_class_any_16 ///

, ///

${graph_opts} ///

over(facility_type) ///

xlabel(${pct}) ///

barlab(mean) ///

legend(rows(1) symxsize(small) symysize(small))

部分选项解释：global graph_opts ///定义暂元

title(, justification(left) ///

color(black) span position(11)) ///前两行为图标题设置

graphregion(color(white)) ///图片背景颜色

ylabel(,angle(0) nogrid) ///Y轴标签设置

xtitle(,placement(left) justification(left)) ///x轴标题设置

yscale(noline) xscale(noline) ///X轴与Y轴刻度设置

legend(region(lcolor(none) fcolor(none))) //图例颜色设置

global pct `' 0 '0%' .25 '25%' .5 '50%' .75 '75%' 1 '100%' ''

quietly do 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/ado/betterbar.ado'

use 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/data/bar-better.dta' , clear

betterbar ///

??_correct checklist refer med_any ///

med_class_any_6 med_class_any_16 ///前两行为变量名

, ///

${graph_opts} ///调用暂元

over(facility_type) ///根据该变量分组

xlabel(${pct}) ///调用暂元

barlab(mean) ///画出各变量均值

legend(rows(1) symxsize(small) symysize(small)) //图例行数与图例标志长宽设置

输出图片如下：

2. 两变量柱状图

代码如下：

global graph_opts1 ///

bgcolor(white) ///

graphregion(color(white)) ///

legend(region(lc(none) fc(none))) ///

ylabel(,angle(0) nogrid) ///

title(, justification(left) color(black) span position(11)) ///

subtitle(, justification(left) color(black))

use 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/data/bar-over.dta' , clear

graph bar treat_correct ///

, ///

over(type) ///

asyvars ///

bargap(20) ///

over(study) ///

over(case) ///

nofill ///

blabel(bar, format(%9.2f)) ///

${graph_opts1} ///

bar(1 , lcolor(black) lwidth(thin) fintensity(100)) ///

bar(2 , lcolor(black) lwidth(thin) fintensity(100)) ///

legend(rows(1) ///

order(0 'Measurement:' 1 'Standardized Patient' 2 'Clinical Vignette')) ///

ytitle('Providers ordering correct treatment {&rarr}', ///

placement(bottom) ///

justification(left)) ///

ylabel(${pct})

部分选项解释：global graph_opts1 ///

bgcolor(white) ///

graphregion(color(white)) ///

legend(region(lc(none) fc(none))) ///

ylabel(,angle(0) nogrid) ///

title(, justification(left) color(black) span position(11)) ///

subtitle(, justification(left) color(black)) //子标题设置

use 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/data/bar-over.dta' , clear

graph bar treat_correct ///

, ///

over(type) ///type包括Standardized Patient与Vignette

asyvars ///将第一个over()设置的变量作为Y轴变量

bargap(20) ///柱之间的距离

over(study) ///study包括Bihar、China与Delhi

over(case) ///case包括Diarrhea (ORS)与Tuberculosis (AFB or CXR)

nofill ///忽略缺少的分类

blabel(bar, format(%9.2f)) ///柱上端的数字设置

${graph_opts1} ///

bar(1 , lcolor(black) lwidth(thin) fintensity(100)) ///柱格式设置

bar(2 , lcolor(black) lwidth(thin) fintensity(100)) ///

legend(rows(1) ///

order(0 'Measurement:' 1 'Standardized Patient' 2 'Clinical Vignette')) ///

ytitle('Providers ordering correct treatment {&rarr}', ///

placement(bottom) ///

justification(left)) ///

ylabel(${pct})

输出图片如下：

3. 两轴组合柱状图

代码如下：

use 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/data/bar-two-axes.dta', clear

* Adjust variable for bar position

gen x = int1mo+.5

* Create individual graphs

* ------------------------

foreach foodGroup in animal fruit grain veg starch processed_sugar {

if '`foodGroup'' == 'animal' local graphTitle Animal Sourced

if '`foodGroup'' == 'fruit' local graphTitle Fruit

if '`foodGroup'' == 'grain' local graphTitle Grain

if '`foodGroup'' == 'veg' local graphTitle Vegetables

if '`foodGroup'' == 'starch' local graphTitle Starchy Foods

if '`foodGroup'' == 'processed_sugar' local graphTitle Processed/Sugar

twoway ///

bar number_group x if food_group=='`foodGroup'', ///

yaxis(1) ytitle('Avg. Number of Foods from' 'Group Consumed Last Month', axis(1)) ///

barwidth(.9) fintensity(inten0) lcolor(black) ///

xlabel(0 '0' 3 '3' 6 '6' 9 '9' 12 '12') ///

ylabel(0 '0' 1 '1' 2 '2' 3 '3', axis(1)) || ///

line total_exp int1mo if food_group=='`foodGroup'', ///

yaxis(2) ytitle('Total Value of Exp.' '1000 Real Tz Sh.', axis(2)) ///

ylabel(0 '0' 500 '500' 1000 '1000' 1500 '1500' 2000 '2000' 2500 '2500', axis(2)) ///

xlabel(3 '3' 6 '6' 9 '9' 12 '12') lwidth(1.2) ///

title('`graphTitle'') xtitle('Month of Interview') ///

graphregion(color(white)) bgcolor(white) ///

legend(off) ///

name('`foodGroup'')

}

* Combine graphs into one

* -----------------------

graph combine starch animal fruit grain processed_sugar veg, ///

graphregion(color(white)) plotregion(color(white))

部分选项解释：use 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/data/bar-two-axes.dta', clear

* Adjust variable for bar position

gen x = int1mo+.5

* Create individual graphs

* ------------------------

foreach foodGroup in animal fruit grain veg starch processed_sugar {

if '`foodGroup'' == 'animal' local graphTitle Animal Sourced

if '`foodGroup'' == 'fruit' local graphTitle Fruit

if '`foodGroup'' == 'grain' local graphTitle Grain

if '`foodGroup'' == 'veg' local graphTitle Vegetables

if '`foodGroup'' == 'starch' local graphTitle Starchy Foods

if '`foodGroup'' == 'processed_sugar' local graphTitle Processed/Sugar

twoway ///

bar number_group x if food_group=='`foodGroup'', ///设置为柱状图

yaxis(1) ytitle('Avg. Number of Foods from' 'Group Consumed Last Month', axis(1)) ///指左边Y轴

barwidth(.9) fintensity(inten0) lcolor(black) ///

xlabel(0 '0' 3 '3' 6 '6' 9 '9' 12 '12') ///

ylabel(0 '0' 1 '1' 2 '2' 3 '3', axis(1)) || ///不同Y轴图形设置之间用“||”隔开

line total_exp int1mo if food_group=='`foodGroup'', ///设置为折线图

yaxis(2) ytitle('Total Value of Exp.' '1000 Real Tz Sh.', axis(2)) ///

ylabel(0 '0' 500 '500' 1000 '1000' 1500 '1500' 2000 '2000' 2500 '2500', axis(2)) ///指右边Y轴

xlabel(3 '3' 6 '6' 9 '9' 12 '12') lwidth(1.2) ///

title('`graphTitle'') xtitle('Month of Interview') ///整体图片设置

graphregion(color(white)) bgcolor(white) ///

legend(off) ///

name('`foodGroup'') //将图片进行临时存储，方便后续组合各图

}

* Combine graphs into one

* -----------------------

graph combine starch animal fruit grain processed_sugar veg, ///需组合的图片名

graphregion(color(white)) plotregion(color(white))

组合图片如下：

4. 并排水平柱状图

代码如下：

global graph_opts ///

note(, justification(left) color(black) span position(7)) ///

title(, justification(left) color(black) span position(11)) ///

subtitle(, justification(left) color(black) span position(11)) ///

graphregion(color(white)) ///

ylabel(,angle(0) nogrid) ///

ytitle('') ///

xtitle(,placement(left) justification(left)) ///

yscale(noline) xscale(noline) xsize(7) ///

legend(region(lcolor(none) fcolor(none)))

quietly do 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/adobetterbar.ado'

quietly do 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/adolabelcollapse.ado'

local n_5 = 599

local n_6 = 601

local title_5 = 'Classic case of presumed TB'

local title_6 = 'TB case with positive sputum report'

quietly forvalues i = 5/6 {

local case = `i' - 4

use 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/data/bar-betterbar.dta', clear

gen n = 1

bysort med_generic: egen med_class_typ = mode(med_class), minmode // Label with mosttypical medicine code

label values med_class_typ med_k

keep if case == `i'

labelcollapse (firstnm) n med_class_typ med_generic_encoded sp_location, ///

by(med_generic facilitycode) ///

vallab(med_class_typ med_generic_encoded sp_location)

labelcollapse (sum) n (firstnm) med_generic_encoded med_class_typ, ///

by(med_generic) ///

vallab(med_class_typ med_generic_encoded)

capture separate n, by(med_generic_encoded) shortlabel

foreach var of varlist n?* {

local theLabel : var label `var'

local theLabel = regexr('`theLabel'','med_generic_encoded == ','')

capture summarize n if med_generic == '`theLabel''

capture local theN = `r(mean)'

label var `var' '`theLabel' [`theN']'

}

foreach var of varlist n?* {

replace `var' = . if `var' < 5 // Exclude low volumes

replace `var' = `var'/`n_`i'' // Number of interactions

quietly summarize `var'

if `r(N)' == 0 drop `var'

}

drop if med_generic == 'Sodium Chloride' // not an active ingredient

betterbar (n?*) , ///

stat(sum) over(med_class_typ) by(med_class_typ) nobylabel nobycolor d(1) ///

legend(span cols(1) position(3) ring(1) symxsize(small) symysize(small) size(small) ///

dropzero ///

xlabel(0 '0%' .2 '20%' .4 '40%' .6 '60%') ///

ysize(6) labsize(2) ///

${graph_opts} ///

title('Case `case' (N=`n_`i'')') subtitle('`title_`i''') ///

name(figure_4_`case')

}

grc1leg figure_4_1 figure_4_2 ///

, ///

position(3) ///

graphregion(color(white)) ///

xsize(7)

部分选项解释：

note()，注释设置

subtitle()，副标题设置

by()与over()搭配使用，使图形能够基于分类清楚展示

grc1leg，将多个图形合并为具有一个共同图例的图形

组合图片如下：

温馨提示： 文中链接在微信中无法生效。请点击底部「阅读原文」。或直接长按/扫描如下二维码，直达原文：

5. 两变量堆积柱状图

代码如下：global graph_opts1 ///

bgcolor(white) graphregion(color(white)) ///

legend(region(lcolor(none) fcolor(none))) ///

ylabel(,angle(0) nogrid) ///

subtitle(, justification(left) color(black) span position(11)) ///

title(, color(black) span)

global pct `' 0 '0%' .25 '25%' .5 '50%' .75 '75%' 1 '100%' ''

use 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/data/bar-stack-by.dta', clear

graph bar med_b2_antister_cat?? ///

if dr_3 == 1 ///

, ///

stack over(checkgroup) nofill ///

ylabel(${pct}) ///

legend(order(5 'No Medication' ///

4 'Antibiotic and Steroid' ///

3 'Antibiotic' ///

2 'Steroid' ///

1 'No Antibiotic or Steroid') ///

cols(1) position(3) ///

symxsize(small) symysize(small) size(small)) ///

${graph_opts1} ///

bar(5, color(white) lcolor(black) lpattern(solid) lwidth(thin)) ///

bar(1,lwidth(thin) lcolor(black)) bar(2,lwidth(thin) lcolor(black)) ///

bar(3,lwidth(thin) lcolor(black)) bar(4,lwidth(thin) lcolor(black)) ///

subtitle('Referral', color(black) justification(center) position(12)) ///

name(figure_1)

graph bar med_b2_antister_cat?? ///

if dr_3 == 0 ///

, ///

stack over(checkgroup) nofill ///

ylabel(${pct}) ///

legend(order(5 'No Medication' ///

4 'Antibiotic and Steroid' ///

3 'Antibiotic' ///

2 'Steroid' ///

1 'No Antibiotic or Steroid') ///

cols(1) position(3) ///

symxsize(small) symysize(small) size(small)) ///

${graph_opts1} ///

bar(5, color(white) lcolor(black) lpattern(solid) lwidth(thin)) ///

bar(1,lwidth(thin) lcolor(black)) ///

bar(2,lwidth(thin) lcolor(black)) ///

bar(3,lwidth(thin) lcolor(black)) ///

bar(4,lwidth(thin) lcolor(black)) ///

subtitle('No Referral', color(black) justification(center) position(12)) ///

name(figure_2)

grc1leg figure_2 figure_1 ///

, ///

position(3) ///

graphregion(color(white)) ///

xsize(7) ///

rows(1) ///

legendfrom(figure_2)

部分选项解释：

stack，表明把变量纵着堆积

grc1leg，rows()，表明两个图的放置位置

grc1leg，legendfrom()，表明使用哪个图的图例

组合图片如下：

6. 并排水平柱状图

代码如下：

global graph_opts ///

title(, justification(left) color(black) span position(11)) ///

graphregion(color(white) lcolor(white) lwidth(med) lalign(center)) ///

ylabel(,angle(0) nogrid) xtitle(,placement(left) justification(left)) ///

yscale(noline) xscale(noline) legend(region(lcolor(none) fcolor(none)))

local opts lwidth(thin) lcolor(white) lalign(center)

label define case 1 'Case 1' 2 'Case 2' 3 'Case 3' 4 'Case 4' , modify

quietly do 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/ado/weightab.ado'

use 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/data/bar-weightab.dta' , clear

weightab ///

correct treat_cxr re_3 re_4 treat_refer t_12 ///

med_any med_l_any_1 med_l_any_2 med_l_any_3 med_k_any_9 ///

if city == 2 ///

[pweight = weight_city] ///

, ///

${graph_opts} ///

barlab ///

barlook(1 `opts' fintensity(100)) ///

title('Patna') ///

over(case) ///

graph ///

legend(off) ///

xlabel(${pct}) ///

name(Fig_1_1)

weightab ///

correct treat_cxr re_3 re_4 treat_refer t_12 ///

med_any med_l_any_1 med_l_any_2 med_l_any_3 med_k_any_9 ///

if city == 3 ///

[pweight = weight_city] ///

, ///

${graph_opts} ///

barlab ///

barlook(1 `opts' fintensity(100)) ///

title('Mumbai') ///

over(case) ///

graph ///

legend(position(5) ring(0) cols(1) symxsize(small) symysize(small)) ///

xlabel(${pct}) ///

name(Fig_1_2)

graph combine Fig_1_1 Fig_1_2 ///

, ///

${comb_opts} ///

xsize(7) r(1)

部分选项解释：

barlook()，柱格式设置

graph，这个选项是命令weightab画图的必选项

组合图片如下：

7. 变量分组、标准误差条和跨组比较的水平柱状图（Horizontal bar plot with grouping of variables, standard error bars, and cross-group comparisons）

代码如下：global pct `' 0 '0%' .25 '25%' .5 '50%' .75 '75%' 1 '100%' ''

quietly do 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/ado/betterbar.ado'

use 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/data/bar-better-ci.dta', clear

betterbar ///

(dr_3 correct_treatment) ///

(med_b2_any_antibiotic med_b2_any_steroid med_b2_any_antister med_l_any_2 med_b2_any_schedule_h med_b2_any_schedule_h1 med_b2_any_schedule_x med_l_any_1) ///

, ///

over(city) xlabel($pct) se bin ///

legend(position(5) ring(0) cols(1) symxsize(small) symysize(small)) ///

ysize(7) n barlab(upper)

部分选项解释：

se，表示添加标准误差条

n，表示在图例中添加每组观测值数量

barlab()，表示数值的显示位置

输出图片如下：

8. 横向堆积柱状图

代码如下：

global graph_opts1 ///

title(, justification(left) color(black) span position(11)) ///

graphregion(color(white) lcolor(white) lalign(center)) ///

ylabwl(,angle(0) nogrid) ///

yscale(noline) legend(region(lcolor(none) fcolor(none)))

global pct `' 0 '0%' .25 '25%' .5 '50%' .75 '75%' 1 '100%' ''

use 'https://gitee.com/arlionn/stata-visual-library/raw/master/Library/data/bar-stack-cat.dta' , clear

local x = 0

quietly foreach var of varlist ///

correct treat_cxr re_3 re_4 treat_refer ///

med_any med_l_any_1 med_l_any_2 med_l_any_3 med_k_any_9 {

mean `var' [pweight = weight_city]

mat a = e(b)

local mean = a[1,1]

local mean = string(round(100*`mean',0))

local mean = substr('`mean'',1,strpos('`mean'','.')+1)

local ++x

local theLabel : var label `var'

local theLabels `' `theLabels' `x' '`theLabel'' '' // [`mean'%]

capture mat drop theResult

reg `var' i.city [pweight = weight_city]

local theR21 = `e(r2)'

mat theResult = nullmat(theResult) , [`theR21']

reg `var' i.city i.case [pweight = weight_city]

local theR22 = `e(r2)' - `theR21'

mat theResult = nullmat(theResult) , [`theR22']

reg `var' i.city i.case i.type_formal [pweight = weight_city]

local theR23 = `e(r2)' - `theR21' - `theR22'

mat theResult = nullmat(theResult) , [`theR23']

reg `var' i.city i.case i.type_formal i.sp_city_id [pweight = weight_city]

local theR24 = `e(r2)' - `theR21' - `theR22' - `theR23'

mat theResult = nullmat(theResult) , [`theR24']

reg `var' i.city i.case i.type_formal i.sp_city_id i.sp_city_mbbs [pweight = weight_city]

local theR25 = `e(r2)' - `theR21' - `theR22' - `theR23' - `theR24'

mat theResult = nullmat(theResult) , [`theR25']

mean `var' [pweight = weight_city]

mat a = e(b)

local mean = a[1,1]

mat theResult = nullmat(theResult) , [`mean']

mat theResults = nullmat(theResults) \ theResult

matlist theResults

}

clear

svmat theResults

gen n = _n

label define n `theLabels'

label values n n

graph bar (sum) theResults1 theResults2 theResults3 theResults4 theResults5 ///

, ///

ylabel($pct) ///

${graph_opts1} ///

horizontal stack over(n) xsize(6) ///

bar(1, lcolor(black) lwidth(thin)) ///

bar(2, lcolor(black) lwidth(thin)) ///

bar(3, lcolor(black) lwidth(thin)) ///

bar(4, fcolor(black) lcolor(black) lwidth(thin)) ///

bar(5, fcolor(gs12) lcolor(black) lwidth(thin)) ///

legend(position(5) ring(0) cols(1) symxsize(small) symysize(small) ///

order(6 'Variance Explained By:' ///

1 'City Setting' ///

2 'Case Scenario' ///

3 'MBBS Degree' ///

4 'All SP Characteristics' ///

5 'Full Interaction Model'))

部分选项解释：

horizontal，表示图形方向设置为横向

输出图片如下：

9. 相关推文

Note：产生如下推文列表的 Stata 命令为：

lianxh 图形 绘图 直方图

安装最新版 lianxh 命令：

ssc install lianxh, replace