编辑 hbar 图 (Stata)

Question

我下面的代码生成了附加的图表。但是，我正在尝试添加两个调整但没有运气。 1- 我想组织 Y 轴，所有行业的 11 月都在 12 月之前，而不是像当前图表中那样按哪个月有更多工作来排列。 2- 我还尝试在 Y 轴上添加标签，它只显示“Nov”和“Dec”，没有额外的文本，虽然 Stata 不会产生任何错误，但它不会改变图表。

preserve
drop if total_jobs_industry<15
graph hbar (count) total_jobs_industry, over(month) over(industry, sort(1)) subtitle("Jobs by Industry and month", span) 
restore 

我知道我可以在 Stata 中手动更改带有微小细节的图表，但如果可能的话，我更喜欢自动化该过程。

数据示例：

 Example generated by -dataex-. To install: ssc install dataex
clear
input float total_jobs_industry str39 industry str8 month
11 "Architectural & Engineering Services" "Nov_2020"
11 "Architectural & Engineering Services" "Nov_2020"
11 "Architectural & Engineering Services" "Dec_2020"
11 "Architectural & Engineering Services" "Dec_2020"
11 "Architectural & Engineering Services" "Nov_2020"
11 "Architectural & Engineering Services" "Dec_2020"
11 "Architectural & Engineering Services" "Dec_2020"
11 "Architectural & Engineering Services" "Nov_2020"
38 "Computer Hardware & Software"         "Dec_2020"
12 "Consulting"                           "Dec_2020"
63 ""                                     "Dec_2020"
32 "IT Services"                          "Dec_2020"
32 "IT Services"                          "Nov_2020"
38 "Computer Hardware & Software"         "Nov_2020"
12 "Aerospace & Defense"                  "Nov_2020"
12 "Accounting"                           "Nov_2020"
12 "Accounting"                           "Dec_2020"

当我使用 sum 而不是 count 时，我得到了下图：

preserve
drop if total_jobs_industry<15
graph hbar (sum) total_jobs_industry, over(month) over(industry, sort(1)) subtitle("Jobs by Industry and month", span) 
restore 

此外，这就是我创建变量以计算每个行业的工作数量的方式：

// The variable id contains observation number running from 1 to X and nt is the total number of observations
generate id = _n
generate nt = _N

// Sorting by inudstry. Now n1 is the observation number within each Industry group and total_jobs_industry is the total number of observations for each Industry group.
sort industry 
by industry: generate n1 = _n
by industry: generate total_jobs_industry = _N
order total_jobs_industry, a(industry)

Answer 1

这是一个非常令人费解的问题。以下原因列表不完整。

1. 该帖子似乎混合了新旧版本，并且不一致。你不能合理地期望我们可靠地解码这样一个曲折的故事。这里的标准是提供一个最小的可验证示例，并且该线程不符合该标准。见guidance here。

2. 显示的图表均不符合给定的数据。

3. 我很难相信(count) 对您的数据有意义。如前所述，它计算非缺失值，但您的关键变量似乎是total_count_industry。另一方面，使用 (sum) 进行各种处理，观察次数似乎会混淆不同类型的计算。

4. 您的示例数据中似乎存在重复的观察结果。

5. 您声明您还尝试在 Y 轴上添加标签，其中仅显示“Nov”和“Dec”，但您的代码中没有显示任何此类评论尝试。

6. 您期望Nov_2020 排在Dec_2020 之前，这不会发生，因为就Stata 而言它只是一个字符串变量，因此D 排在N 之前这一事实至关重要.这就是 12 月在 1 月之前排序的原因，这与按行业值排序无关，它只影响条形组的排序。您没有使用 Stata 的日期变量功能。

我怀疑除了最后一个问题，我是否能理解这些问题中的任何一个。 graph hbar 的限制似乎是它忽略了时间变量的显示格式，所以我使用值标签来确保 Nov 和 Dec 按您希望的顺序排序。

clear
input float total_jobs_industry str39 industry str8 month
11 "Architectural & Engineering Services" "Nov_2020"
11 "Architectural & Engineering Services" "Nov_2020"
11 "Architectural & Engineering Services" "Dec_2020"
11 "Architectural & Engineering Services" "Dec_2020"
11 "Architectural & Engineering Services" "Nov_2020"
11 "Architectural & Engineering Services" "Dec_2020"
11 "Architectural & Engineering Services" "Dec_2020"
11 "Architectural & Engineering Services" "Nov_2020"
38 "Computer Hardware & Software"         "Dec_2020"
12 "Consulting"                           "Dec_2020"
63 ""                                     "Dec_2020"
32 "IT Services"                          "Dec_2020"
32 "IT Services"                          "Nov_2020"
38 "Computer Hardware & Software"         "Nov_2020"
12 "Aerospace & Defense"                  "Nov_2020"
12 "Accounting"                           "Nov_2020"
12 "Accounting"                           "Dec_2020"
end 

duplicates drop 

gen mdate = monthly(month, "MY")

levelsof mdate, local(months)
tokenize "`c(Mons)'" 
foreach m of local months { 
    local month = month(dofm(`m'))
    label def mdate `m' "``month''", modify 
}
label val mdate mdate 

set scheme s1color 
graph hbar (asis) total_jobs_industry, over(mdate) over(industry, sort(1) descending)