stata备忘录

阅读本书更多章节>>>>

Stata备忘录

1. 画图

（1）时间趋势图

label var year "年份"label var per "制造业增加值比重[左轴]"label var tjj "工业增加值比重[右轴]"graph twoway (connect per year ,yaxis(1) color(black) )   ///(connect tjj year ,yaxis(2) color(black) lpattern(dash)  ) ///, graphregion(color(white))  xlabel(2003(2)2019) ///ytitle("世界银行制造业增加值比重(%)",axis(1) height(5))  ///ytitle("国家统计局工业增加值比重(%)",axis(2) height(5))  ///note(数据来源：World Bank Open Data、国家统计局) xline(2011) 

等价命令

tw (connect value1819 season , /// lcolor(black) lpattern(dash) msymbol(O) mlcolor(gs5) mfcolor(gs12)) /// (connect value2020 season , ///lcolor(black) lpattern(solid) msymbol(S) mlcolor(gs5) mfcolor(gs12)) ///,graphregion(color(white)) ///legend(label(1 "18-19年平均") label(2 "2020年") ) ///xlabel(1 "第一季度"  2 "第二季度"  3 "第三季度"  4 "第四季度"  ,labsize(small) )

label var year "年份"tw bar mR1 year,yaxis(2) bc(balck) sort barwidth(0.9) fintensity(inten0) ///ylabel(0(2000)6000, axis(2)) /// xlabel(2014(1)2021)|| /// connect percent_R year,yaxis(1) lc(black) lp(dash) mc(blace) ///ylabel(0.5 "50%" 0.6 "60%" 0.7 "70%" 0.8 "80%" ,axis(1))  ||, ///graphregion(color(white) ) ///bgcolor(white) ///title("中国数字内容企业(游戏)收入金额及占全球市场比重", c(black) size(*0.8)) ///ytitle("占比（%）",axis(1) height(7))  ///ytitle("收入额（百万美元）",axis(2) height(5))  /// legend(label(1 "中国数字内容企业(游戏)收入占全球市场比重") label(2 "中国数字内容企业(游戏)收入金额") ) ///legend(size(small) col(1)) ///note("数据源自：app annie")graph save "Graph" "$path\output\playdata_1_percent_and_value_of_Chinese_Apps_Export.gph",replace

use hs_adj_year_PQV_2000_2015.dta,clearuse hs_adj_year_PQV_2000_2015_cregime10,clearmerge m:1 hs_adj using equipmentreplace BEC=4 if BEC==1 & equipment!=1destring hs_adj,replacereghdfe lnV i.year if year!=2006 & BEC==2, a(hs)est store result_accessoriesreghdfe lnV i.year if year!=2006 & BEC==4, a(hs)est store result_equipmentreghdfe lnV i.year if year!=2006 & BEC==0, a(hs)est store result_noncapital#d ;coefplot   (result_accessories,c(l) label("accessories") lp(dash) lc(black) mc(black) ms(smcircle_hollow) offset(-0.07))    (result_equipment ,c(l)  label("equipment") lp(solid) lc(black) mc(black) ms(smcircle_hollow))  (result_noncapital,c(l)  label("noncapital") lp(dot) lc(black) mc(black) ms(smcircle_hollow) offset(0.07))  , vertical   drop(_cons) byopts(xrescale)   xlabel(1 "2001"  3"2003"  5"2005" 6"2007"  8"2009"  10"2011" 12"2013" 14"2015")  graphregion(color(white))  legend(size(small) col(3))  ;#d cr

字体大小 option

字体大小option	description
zero	no size whatsoever, vanishingly small
minuscule	smallest
quarter_tiny
third_tiny
half_tiny
tiny
vsmall
small
medsmall
medium
medlarge
large
vlarge
huge
vhuge	largest
tenth	one-tenth the size of the graph
quarter	one-fourth the size of the graph
third	one-third the size of the graph
half	one-half the size of the graph
full	text the size of the graph
size	any size you want

节点样式 eg: msymbol(O) mlcolor(gs5) mfcolor(gs12)

symbolstyle	Synonym(if any)	Description
circle	O	solid
diamond	D	solid
triangle	T	solid
square	S	solid
plus	+
X	X
arrowf	A	filled arrow head
arrow	a
pipe
V	V
smcircle	o	solid
smdiamond	d	solid
smsquare	s	solid
smtriangle	t	solid
smplus
smx	x
smv	v
circle_hollow	Oh	hollow
diamond_hollow	Dh	hollow
triangle_hollow	Th	hollow
square_hollow	Sh	hollow
smcircle_hollow	oh	hollow
smdiamond_hollow	dh	hollow
smtriangle_hollow	th	hollow
smsquare_hollow	sh	hollow
point	p	a small dot
none	i	a symbol that is invisible

线样式

linepatternstyle	Description
solid	solid line
dash	dashed line
dot	dotted line
dash_dot
shortdash
shortdash_dot
longdash
longdash_dot
blank	invisible line
formula	e.g.,-. or --… etc.
A formula is composed of any combination of
l	solid line
_	(underscore) a long dash
-	(hyphen) a medium dash
.	short dash (almost a dot)
#	small amount of blank space

颜色

black	edkblue	gs12	lime	orange
blue	eggshell	gs13	ltblue	orange_red
bluishgray	eltblue	gs14	ltbluishgray	pink
bluishgray8	eltgreen	gs15	ltbluishgray8	purple
brown	emerald	gs16	ltkhaki	red
chocolate	emidblue	gs2	magenta	sand
cranberry	erose	gs3	maroon	sandb
cyan	forest_green	gs4	midblue	sienna
dimgray	gold	gs5	midgreen	stone
dkgreen	gray	gs6	mint	sunflowerlime
dknavy	green	gs7	navy	teal
dkorange	gs0	gs8	navy8	white
ebblue	gs1	gs9	none	yellow
ebg	gs10	khaki	olive
edkbg	gs11	lavender	olive_teal

（2）柱状图

#delimit ;graph bar cn_wzje_80 cn_wzje_81 if wzlx==0, over(sec) bargap(-30)  ytitle("吸引外资金额")  legend( label(1 "08年前") label(2 "08年后") )  title("各行业吸引外资2008年前后对比")  subtitle("中国中西部服务业")  note("中国中西部服务业") ;#delimit cr

（3）散点图

tw (scatter delta_v2_v3 delta_v1_v2 if delta_v1_v2>=-0.3& delta_v1_v2 <=2.3 ///&delta_v2_v3>= -0.3&delta_v2_v3<=2.3, ///mlabel(hy4) mlc(black) mlabc(black) ms(x) mlabs(tiny)) ///(fun y=x,range(-0.3 2.3)) , ///xlab(-0.3(0.5)2.3) ylab(-0.3(0.5)2.3)  ///graphregion(color(white)) ///xline(0,lp(dash) lc(gs10)) ///yline(0,lp(dash) lc(gs10)) ///legend(ring(0) pos(5) order(2 "45°线")) ///ytitle("11-15时段的增速") ///xtitle("07-11时段的增速")

graph twoway (scatter wzje_CE0308 wzje_CE1419, mlabel(sec) mlabv(sec) ) (function y=x, range(0 0.11)) , ///title("中国东部地区03-08对比中国东部地区14-19[金额]")  ///ytitle("中国东部03-08") ///xtitle("中国东部14-19") ///legend(ring(0) pos(5) order(2 "45°线")) ///graphregion(color(white))

graph twoway (scatter c_AS c_WD, mlabel(cic03)  ) (lfit c_AS c_WD) , ///title("东盟增速放缓 vs 世界增速放缓")  ///ytitle("东盟")  ///xtitle("世界") ///legend(ring(0) pos(5) order(2 "拟合")) ///graphregion(color(white))

（4）bgshade

bgshade ks, shaders(uu9)  ///   twoway(connect lamda22 ks if treat==1&ks>=6&ks<=11 ||             ///            connect lamda22 ks if treat==0&ks>=6&ks<=11 , xlab(6(1)11) ///   title("新冠疫情冲击下企业平均收入变化趋势"))

（5）coefplot

coefplot,  levels(90) vertical lcolor(black)mcolor(black) /// msymbol(circle_hollow) ytitle(估计系数, size(small)) /// ylabel(, labsize(small) angle(horizontal) nogrid) /// yline(0, lwidth(vthin)lpattern(solid) lcolor(black)) /// xtitle(事件发生时间, size(small)) ///  title("(B)企业缴税的平行趋势检验") /// xlabel(0"." 1"2019s2" 2"2019s3" 3"2019s4" 4"2020s1" 5"2020s2" 6"2020s3" 7"2020s4") 

reghdfe lnQ i.Year if elec == 1,a(i.citycode) vce(r)est store elec_Q_1 reghdfe lnV i.Year if elec == 1,a(i.citycode) vce(r)est store elec_V_1 reghdfe lnQ i.Year if elec == 0,a(i.citycode) vce(r)est store elec_Q_0reghdfe lnV i.Year if elec == 0,a(i.citycode) vce(r)est store elec_V_0coefplot (elec_Q_1,label("半导体电子元件相关企业进口数量") offset(0.05)  pstyle(p3)) ///(elec_Q_0 ,label("非半导体电子元件相关企业进口数量") offset(-0.05)  pstyle(p4) ), ///vertical drop(_cons) xline(0) ///graphregion(color(white)) /// yline(0) ///addplot(line @b @at,lp(dash) lwidth(*0.5)) /// legend(label(1 "半导体电子元件相关企业进口数量") label(2 "非半导体电子元件相关企业进口数量") ) 

（6）画系数和置信区间

twoway (scatter coef week) /// (rcap ci_lower ci_upper week, ///  lcolor(black) ///  mcolor(black) ///  lwidth(vthin) ///  lpattern(dash) /// msymbol(circle_hollow) /// legend(label(2 "99% CI"))) , ///yline(0) ///xtitle("")  ///graphregion(fcolor(white)) ///title("第X周的系数", size(medium)) /// name("Coef_all_I", replace)

（7）画直方图

一般使用kdensity

hist year if year>=1400 & year<=2010, freq bin(200) ylabel(0(500)2500) xtitle("Year") xline(1950 1980,lw(thin)) ///     text(1500 1950 "Year=1950", place(w)) text(2000 1980 "Year=1980", place(w)) 

（8）画桑基图

cd $path\appdatause Data_games.dta,clearmerge m:1 ParentCompanyName using "$path\data\company_city"keep if _m == 3drop _mgen from = city_codeencode iso3_j,gen(to)bys from to :egen tR = total(Revenue)bys from to :egen tD = total(Downloads)duplicates drop  from to ,forcegen x0 = 1gen x1 = 2tostring city_code ,gen(city2)drop if dest == "CHN"sankey_plot x0 from x1 to, ///width0(tR) extra xlabel(1 "Source" 2 "Destination", nogrid labsize(small)) ///colorpalette(economist, opacity(30)) ///label0(city) label1(iso3_j) ///labsize(*0.6) labcolor(black) ///graphregion(color(white))  gap(0.1) ///title("地级市层面Apps出海流向（按收入额）",color(black) size(*0.8))graph save "Graph" "$path\output\sankey_R_0228.gph",replacesankey_plot x0 from x1 to, ///width0(tD) extra xlabel(1 "Source" 2 "Destination", nogrid labsize(small)) ///colorpalette(economist, opacity(30)) ///label0(city) label1(iso3_j) ///labsize(*0.6) labcolor(black) ///graphregion(color(white))  gap(0.1) ///title("地级市层面Apps出海流向（按下载量）",color(black) size(*0.8))graph save "Graph" "$path\output\sankey_D_0228.gph",replace

（9）气泡图

twoway(scatter mv T_gap_05_00 [fweight=N] if BEC == 0&T_gap_05_00!=0&N !=0,msymbol(Oh) mc(ebblue%40)) ///(scatter mv T_gap_05_00 [fweight=N] if BEC == 4&T_gap_05_00!=0&N !=0,msymbol(Oh) mc(orange_red%40)) ///(scatter mv T_gap_05_00 [fweight=N] if BEC == 2&T_gap_05_00!=0&N !=0,msymbol(Oh) mc(green%20)) ///, legend(label(1 "非资本品") label(2 "equipment") label(3 "accessories") ) 

2. 处理数据

（1）拓展expand数据

例如：当前数据中 有9523、9524、9525、9526、9527、9528共计6个样本，现在想把这6个样本根据freq进行扩充，变为9523、9524_1、9524_2、9525_1、9525_2、9526_1、9526_2、9527_1、9527_2、9528_1、9528_2这11个样本

freq	count	value
1	9523	4845.1143
2	9524	969.66498
2	9525	129.53349
2	9526	71284.508
2	9527	1038.127
2	9528	445877.09

count是id的唯一识别码，expandcl函数可以生成freq行相同的样本，并生成一个新的id识别码freq_count

egen count=group(id hs02_6)expandcl freq,gen(freq_count) cluster(count)drop freq_count

（2）时间数据

gen R= mdy(month_r,day_r,year_r)gen week_r = week(R)gen day_r = day(R)gen dow_r = dow(R)  //返回周几gen doy_r = doy(R)  //返回年内日期gen yw_r = yw(year_r,week_r)gen ed = yw - yw_r//yw ym yq yh分别为年周、年月、年季、年半年

gen period_kb= date(date_u,"YMD")-date(date_kb,"YMD")

（3）常见函数

int(x) //取整，不论后面的小数是什么，只取小数点前的数值round(x) // 四舍五入取整round(x, .01) //保留两位小数四舍五入gen y = sum(x)  //求列累积和egen y = sum(x) //求列总和egen y = rsum(x y z) //求x+y+z总和egen y = rowmean(x y z) //求(x+y+z)/3egen y = rowsd(x y z) //求x y z的方差egen y = rowmim(x y z) //求x y z的最小值egen y = rowmax(x y z) //求x y z的最大值egen y = mean(x)      //求列均值egen y = median(x)    //求列中位数egen y = std(x)       //求变异系数，与方差不同bysort x(y): gen z = y[1] //按照x分组，分组后按照y排序，生成一个新变量z=y的第一个观察值

（4）缩尾处理

foreach v of var DexpoAS4- DlnexpoWD2{gen `v'_w=`v'qui su `v',detreplace `v'_w=r(p99) if `v'>r(p99) & `v'<.replace `v'_w=r(p1) if `v'<r(p1)}winsor2 wage, replace cuts(1 99) trim

summary 一个变量之后，可以返回的结果有

r(N)           //number of observationsr(mean)        //meanr(skewness)    //skewness (detail only)r(min)         //minimumr(max)         //maximumr(sum_w)       //sum of the weightsr(p1)          //1st percentile (detail only)r(p5)          //5th percentile (detail only)r(p10)         //10th percentile (detail only)r(p25)         //25th percentile (detail only)r(p50)         //50th percentile (detail only)r(p75)         //75th percentile (detail only)r(p90)         //90th percentile (detail only)r(p95)         //95th percentile (detail only)r(p99)         //99th percentile (detail only)r(Var)         //variancer(kurtosis)    //kurtosis (detail only)r(sum)         //sum of variabler(sd)          //standard deviation

（5）创建文件夹

在project路径下生成一个workspace文件夹
再生成子文件夹储存数据（data）、控制变量（controlvars ）、临时数据（tempdata）、回归结果（outreg）

efolder, cd(D:\stata15\project\workspace)efolder, cd(D:\stata15\project\workspace sub(data controlvars tempdata outreg)

（6）分组处理数据（bysort的替代方案）

*展示根据highzupu50（族谱）和year分组后的变量drqianfen(死亡率)均值；collapse (mean) drqianfen, by(highzupu50 year)

（7）定义无缺失的样本

g rsample = !mi(avggrain_fyr) & !mi(nograin_fyr) & !mi(urban_fyr)& !mi(dis_bj_fyr) & !mi(dis_pc_fyr) & !mi(migrants_fyr)& !mi(rice_fyr) & !mi(minor_fyr) & !mi(edu_fyr)

（8）定义Dummy的新替代式（时间range）

*如果yob满足1825≤yob≤1899则pre取值为1,否则pre取值为0。mid、post生成过程类似。gen pre = inrange(yob, 1825, 1899)gen mid = inrange(yob, 1899, 1919)gen post = inrange(yob, 1920, 1960)

（9）快速替换

recode treatyear (1969 = 1) (1979 = 2) (1989 = 3) (1999 = 4) (2009 = 5)

3. 处理字符

（1）替换字符

replace 候选人姓名=subinstr( 候选人姓名, " ", "",. )

（2）捕捉字符中的某些特征

keep if strmatch(city, "*山东*")gen temp = 1 if strmatch(reporteriso3, "A*")

（3）提取字符，检索特定字符

//从enddate字符1开始取，取4个字符赋给yeargen year = substr(enddate,1,4)  //strpos(s1, s2)返回字符s2在s1中的位置，如果s1中找不到s2，则返回0，将该判断再赋给ygen y = strpos(s1, s2) != 0    

4. 输出结果

（1）常规输出

outreg2 using "E:\mfg\outreg\r2", word append addtext(CountryFE, YES,YearFE, YES)

（2）iv回归输出第一阶段

eststo: xtivreg p_a_w (DexpoCN4_w=dexpo44) ///  i.t c.expr0#t c.Lshare0#t ///  c.lnGDP0#t c.lngdp0#t, fe first vce(cluster c)    eststo: xtreg DexpoCN4_w dexpo44 ///  i.t c.expr0#t c.Lshare0#t ///  c.lnGDP0#t c.lngdp0#t if e(sample)==1 ,fe   cd $path\outreg  outreg2 using "table3", word replace addtext(CityFE, YES,YearFE, YES) keep(dexpo44) 

（3）变量描述性统计

*列出inv等变量的样本数、均值、标准差、最小值和最大值。tabstat inv loginv log_levies ///     logpopl logincome logasset hhsize landpc logmigration logtax logtransfer share_admin  /// postcont postopen secret_ballot proxy_voting moving_ballot /// , s(N mean sd min max) c(s)

5.矩阵保存结果

mat T1 = J(3,3,.)reghdfe temp ib1.season if year == 2018 & treat == 1,noaforvalues i = 1/3{local j = `i' + 1mat T1[`i',1] = _b[`j'.season]}reghdfe temp ib1.season if year == 2019 & treat == 1,noaforvalues i = 1/3{local j = `i' + 1mat T1[`i',2] = _b[`j'.season]}reghdfe temp ib1.season if year == 2020 & treat == 1,noaforvalues i = 1/3{mat T1[`i',3] = _b[`=1+`i''.season]}svmat T1

6.导入数据（全字符串）

forv i = 2000/2003{cd E:\Data\EPS工企海关匹配库\origindataimport delimited "工企+海关(`i').csv", stringcols(_all) clear cd E:\Data\EPS工企海关匹配库save data`i'.dta,replace}

7.循环

clear allset obs 1000**#** 用forvalues循环对单一变量进行处理gen id = .    //生成一个变量名为id，代表第几个人，假设一共有50个人//假设每个人都有20个观测值，代表20年forvalues i = 1/50 {local j = `i' - 1//暂时定义0~49 方便计算local lower = `j' * 20  +1//定义下限 1、21、41、61 local upper = `j' * 20 + 20//定义上限 20、40、60、80//由此就定义了 1~20  21~40 41~60 ……replace id = `i' in `lower'/`upper'//给第1~20行，赋值为第1个人//给第21~40行，赋值为第2个人}bys id : gen T = _n + 2000//对于每个人，都生成一个时间序列**#** 用forvalues循环对多个变量进行处理forvalues i = 1/5 {gen value`i' = .cap gen e = rnormal()replace value`i' = e * 10 + `i'cap drop e}// 等价于 gen value6 = .cap gen e = rnormal()replace value6 = e * 10 + 6cap drop egen value7 = .cap gen e = rnormal()replace value7 = e * 10 + 7cap drop egen value8 = .cap gen e = rnormal()replace value8 = e * 10 + 8cap drop egen value9 = .cap gen e = rnormal()replace value9 = e * 10 + 9cap drop egen value10 = .cap gen e = rnormal()replace value10 = e * 10 + 10cap drop e**#** 用while循环对单一变量进行处理// 只要时间在T=11和T=20之间，就对value1~value10进行 " 乘0.1"的处理local i = 2010 while `i' < 2020 {forvalues j = 1/10{replace value`j' = value`j' * 0.1 if T == `i'}local i = `i' + 1}**#** 用foreach对变量进行处理foreach v in value1 value2 value3 value4 value5  {su `v' ,dreplace `v' = (`v' - r(min)) / (r(max) - r(min))kdensity `v'}// 等价于 su value6,d replace value6 = (value6 - r(min)) / (r(max) - r(min))kdensity value6su value7,d replace value7 = (value7 - r(min)) / (r(max) - r(min))kdensity value7su value8,d replace value8 = (value8 - r(min)) / (r(max) - r(min))kdensity value8su value9,d replace value9 = (value9 - r(min)) / (r(max) - r(min))kdensity value9su value10,d replace value10 = (value10 - r(min)) / (r(max) - r(min))kdensity value10

8.将第一行作为 label / varname

* 把第一行作为变量标签labone,nrow(1) * 把第一行作为变量名（不保留第一行）nrow* 把第一行作为变量名（保留第一行）nrow,keep

阅读本书更多章节>>>>

本文链接：https://www.kjpai.cn/gushi/2024-04-24/161964.html，文章来源：网络cs，作者：康由，版权归作者所有，如需转载请注明来源和作者，否则将追究法律责任！