计算当前值等于或小于前 90 天值的次数

Question

我有一个每日交易数据集，其中存在差距。我想查看在日期 T 参考价格 websiteprice 为 $X 的产品是否在前 T – 90 天的至少 10% 内具有实际售价 actualsoldprice $Y >= $X。换句话说，对于每笔 sale_at_or_above_refprice == 1 的交易，我们需要计算过去 90 天内（给定产品）先前交易的实际售价达到或超过该交易参考价格的次数。

我在 wanted 变量中包含了我要查找的第一步结果。

我的数据如下，

* Example generated by -dataex-. For more info, type help dataex
clear
input str9 orderdate str16 productcode str10 productcategory byte(websiteprice actualsoldprice sale_at_or_above_refprice var7 wanted)
"3-Jan-20"  "MZZ32819-564-282" "Mens Jeans" 40 25 . .  .
"8-Jan-20"  "MZZ32819-564-282" "Mens Jeans" 40 40 1 .  .
"12-Jan-20" "MZZ32819-564-282" "Mens Jeans" 40 40 1 .  1
"12-Sep-20" "MZZ32819-564-282" "Mens Jeans" 40 28 . .  .
"18-Sep-20" "MZZ32819-564-282" "Mens Jeans" 40 24 . .  .
"20-Sep-20" "MZZ32819-564-282" "Mens Jeans" 50 30 . .  .
"27-Sep-20" "MZZ32819-564-282" "Mens Jeans" 50 25 . .  .
"11-Oct-20" "MZZ32819-564-282" "Mens Jeans" 40 20 . .  .
"19-Oct-20" "MZZ32819-564-282" "Mens Jeans" 35 24 . .  .
"2-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 20 20 1 .  6
"2-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 14 14 1 .  7
"4-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 14 14 1 .  8
"7-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 14 14 1 .  9
"7-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 20 20 1 .  7
"9-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 20 20 1 .  8
"11-Nov-20" "MZZ32819-564-282" "Mens Jeans" 14 14 1 . 12
"12-Nov-20" "MZZ32819-564-282" "Mens Jeans" 14 14 1 . 13
"14-Nov-20" "MZZ32819-564-282" "Mens Jeans" 14 14 1 . 14
"15-Nov-20" "MZZ32819-564-282" "Mens Jeans" 14 14 1 . 15
"18-Nov-20" "MZZ32819-564-282" "Mens Jeans" 14 14 1 . 16
"24-Nov-20" "MZZ32819-564-282" "Mens Jeans" 20 20 1 .  9
end

编辑 - 我已经更新了 wanted 变量并包含了一个 new_wanted。不同之处在于考虑了具有多个价格的重复日期。还包括 2 个产品到运行这个过程的 id。

* Example generated by -dataex-. For more info, type help dataex
clear
input str9 orderdate str16 productcode str10 productcategory byte(websiteprice actualsoldprice sale_at_or_above_refprice wanted new_wanted)
"3-Jan-20"  "MZZ32819-564-282" "Mens Jeans" 40 25 .  .  .
"8-Jan-20"  "MZZ32819-564-282" "Mens Jeans" 40 40 1  0  0
"12-Jan-20" "MZZ32819-564-282" "Mens Jeans" 40 40 1  1  1
"12-Sep-20" "MZZ32819-564-282" "Mens Jeans" 40 28 .  .  .
"18-Sep-20" "MZZ32819-564-282" "Mens Jeans" 40 24 .  .  .
"20-Sep-20" "MZZ32819-564-282" "Mens Jeans" 50 30 .  .  .
"27-Sep-20" "MZZ32819-564-282" "Mens Jeans" 50 25 .  .  .
"11-Oct-20" "MZZ32819-564-282" "Mens Jeans" 40 20 .  .  .
"19-Oct-20" "MZZ32819-564-282" "Mens Jeans" 35 24 .  .  .
"2-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 20 20 1  6  6
"2-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 14 14 1  6  6
"4-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 14 14 1  8  7
"7-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 14 14 1  9  8
"7-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 20 20 1  7  7
"9-Nov-20"  "MZZ32819-564-282" "Mens Jeans" 20 20 1  8  9
"11-Nov-20" "MZZ32819-564-282" "Mens Jeans" 14 14 1 12 10
"12-Nov-20" "MZZ32819-564-282" "Mens Jeans" 14 14 1 13 11
"14-Nov-20" "MZZ32819-564-282" "Mens Jeans" 14 14 1 14 12
"15-Nov-20" "MZZ32819-564-282" "Mens Jeans" 14 14 1 15 13
"18-Nov-20" "MZZ32819-564-282" "Mens Jeans" 14 14 1 16 14
"24-Nov-20" "MZZ32819-564-282" "Mens Jeans" 20 20 1  9  9
"6-Jan-20"  "ADDZ4449-524-645" "Mens Bags"  60 50 .  .  .
"11-Jan-20" "ADDZ4449-524-645" "Mens Bags"  70 60 .  .  .
"12-Feb-20" "ADDZ4449-524-645" "Mens Bags"  60 60 1  .  1
"12-Jul-20" "ADDZ4449-524-645" "Mens Bags"  60 50 .  .  .
"18-Sep-20" "ADDZ4449-524-645" "Mens Bags"  50 55 1  .  1
"20-Sep-20" "ADDZ4449-524-645" "Mens Bags"  50 45 .  .  .
"20-Sep-20" "ADDZ4449-524-645" "Mens Bags"  66 45 .  .  .
"12-Oct-20" "ADDZ4449-524-645" "Mens Bags"  55 60 1  .  1
"19-Oct-20" "ADDZ4449-524-645" "Mens Bags"  60 60 1  .  1
"2-Nov-20"  "ADDZ4449-524-645" "Mens Bags"  70 73 1  .  0
"2-Nov-20"  "ADDZ4449-524-645" "Mens Bags"  60 56 .  .  .
"4-Nov-20"  "ADDZ4449-524-645" "Mens Bags"  60 60 1  .  3
"7-Nov-20"  "ADDZ4449-524-645" "Mens Bags"  50 45 .  .  .
"7-Nov-20"  "ADDZ4449-524-645" "Mens Bags"  66 66 1  .  1
"9-Nov-20"  "ADDZ4449-524-645" "Mens Bags"  60 56 .  .  .
"11-Nov-20" "ADDZ4449-524-645" "Mens Bags"  60 76 1  .  5
"12-Nov-20" "ADDZ4449-524-645" "Mens Bags"  60 71 1  .  6
"13-Nov-20" "ADDZ4449-524-645" "Mens Bags"  60 26 .  .  .
"15-Nov-20" "ADDZ4449-524-645" "Mens Bags"  65 70 1  .  4
"15-Nov-20" "ADDZ4449-524-645" "Mens Bags"  67 70 1  .  3
"22-Nov-20" "ADDZ4449-524-645" "Mens Bags"  56 70 1  .  9
end

下面是我试图为这个任务改编的代码。感谢 STATALIST 的 Ken Chui。

gen date1 = date(orderdate, "DMY", 2020)
format date1 %td

local max = _N
gen wanted2 = .
foreach x of numlist 1/`max'{
    capture drop get get_sum
    gen get = actualsoldprice >= actualsoldprice[`x']
    rangestat (sum) get, interval(date -90 -1)
    replace wanted2 = get_sum if _n == `x'
}
replace wanted2 = . if sale_at_or_above_refprice == .

Answer 1

*Start by converting date to Stata date
gen stata_date = date(orderdate,"DM20Y")
format stata_date %td

*Sort data and product code as stop conditions in while loop expect them to be sorted
sort productcode stata_date

*Create varialbe to store result
gen count_less = .

*Loop over all rows
count 
forvalue row = 1/`r(N)' {
    
    *Only applicable to 
    if sale_at_or_above_refprice[`row'] == 1 {
        
        *Set result variable to 0 for this row
        replace count_less = 0 if _n == `row'
        
        *Initate locals used in while loop
        local true = 1
        local row_skip = 1
        local count = 0
        local last_date = stata_date[`row']
        
        *Loop until any stop condition sets local true to 0
        while `true' == 1 {
           
            *Test if row_skip hits top of data set (i.e row 0)
            if `row'-`row_skip' == 0                                        local true = 0
            *Test that product is same in compare row
            else if productcode[`row'] != productcode[`row'-`row_skip']     local true = 0
            *Test that previous order is within 90 days
            else if stata_date[`row'] - stata_date[`row'-`row_skip'] > 90   local true = 0

            *Test if actualsoldprice is less thatn old websiteprice
            else if websiteprice[`row'] <= actualsoldprice[`row'-`row_skip'] {
                
                * Each date can only be counted once, so test if date is last date counted 
                if `last_date' != stata_date[`row'-`row_skip'] {
                    *Compare row fits condition, add 1 to counter
                    local count = `count' + 1   
                    
                    *Update last counted date
                    local last_date = stata_date[`row'-`row_skip']
                }
            }
            *Skip one more prevuous row
            local row_skip = `row_skip' + 1
        }
        *Add the count result to the result varaible for this row
        replace count_less = `count' if _n == `row'
    }
}

计算当前值等于或小于前 90 天值的次数

Counting occurrences where current value is equal to or less than previous 90 day values

stata