在上過 David Ko 的 Kanban 課後，有學到一個有趣實用的圖表，就是累積流程圖（Cumulative Flow Diagram），本篇文章會著重在實作的部分，詳細概念的部分可以先參照 Ruddy Lee 老師分享的〈累積流程圖 Cumulative Flow Diagram 解密〉一文。

圖源：The Cumulative Flow Chart (CFD) in a nutshell – Leanguru

這個圖表的製法，可以說是要為 Product Backlog 每隔一段時間進行一次快照（snapshot)，把當時各狀態有多少的 Item 給揭露出來。

但因為 Jira 只能查到當下的狀態，沒辦法溯源。所以這就仰賴我們前面在設計 Issue Type 時留下的各種時間戳（timestamp）。目前有留下的時間戳有：

• Created At
• Refined At
• Sprint began at
• Work began at
• Released at

透過這些時間戳，我們就能計算出特定時間點的以下狀態：

• To Do：在 Product Backlog 中，想要做但還沒有 Refine 的數量
• To Arrange：在 Product Backlog 中，已經 Refine 並等待進 Sprint 開發的數量
• To Develop：已經放進 Sprint，等待開發的數量
• Developing：在 Sprint 且已經展開開發的數量
• Released：已經釋出給使用者的的數量

這些狀態分別可以用以下虛擬碼去理解怎麼算出來的：

To Do

COUNT(
  FILTER("
    `created at` <= theDate AND
    (`refined at` = EMPTY || `refined at` > theDate)
  ")
)

To Arrange

COUNT(
  FILTER("
    `refined at` <= theDate AND
    (`sprint began at` = EMPTY || `sprint began at` > theDate)
  ")
)

To Develop

COUNT(
  FILTER("
    `sprint began at` <= theDate AND
    (`work began at` = EMPTY || `work began at` > theDate)
  ")
)

Developing

COUNT(
  FILTER("
    `work began at` <= theDate AND
    (`released at` = EMPTY || `released at` > theDate)
  ")
)

Released

COUNT(
  FILTER("
    `released at` <= theDate
  ")
)

假設透過 Jira 下載的試算表名為 PBI，並有以下結構：

|     | A   | B       | C          | D               | F             | G           | H        | I      |
| --- | --- | ------- | ---------- | --------------- | ------------- | ----------- | -------- | ------ |
| 1   | Key | Created | Refined at | Sprint began at | Work began at | Released at | Resolved | parent |
| 2   | ... | ...     | ...        | ...             | ...           | ...         | ...      | ...    |
| 3   | ... | ...     | ...        | ...             | ...           | ...         | ...      | ...    |
| 4   | ... | ...     | ...        | ...             | ...           | ...         | ...      | ...    |

並另外有一表名為 CFD，並有以下結構：

|     | A    | B     | C          | D          | F          | G        |
| --- | ---- | ----- | ---------- | ---------- | ---------- | -------- |
| 1   | Date | To Do | To Arrange | To Develop | Developing | Released |
| 2   | ...  | ...   | ...        | ...        | ...        | ...      |
| 3   | ...  | ...   | ...        | ...        | ...        | ...      |
| 4   | ...  | ...   | ...        | ...        | ...        | ...      |
| ... | ...  | ...   | ...        | ...        | ...        | ...      |
| 10  | ...  | ...   | ...        | ...        | ...        | ...      |

其中 B2 的 Google Sheet 公式即為下方程式碼區塊所示，B2 - G10 可依此類對辦理。

=COUNTA(IFNA(FILTER('PBI'!$A$2:$A, 'PBI'!$B$2:$B <= A2, 'PBI'!C2:C > A2), ))

透過這樣的公式這樣我們就能建立一個如下圖的表格與圖表。表裡的數字是我先隨意填寫的，不是真實存在的數字，但若套用在現有紀錄時間超過數週的 Jira Project，會產出有意的表格，而圖表則是累積流量圖。

這樣就成功建立了累積流程圖，讓我們能夠藉由這個圖去做對應的檢視與調適。