如何在子查询中优化SELECT COUNT

Question

让我说我有事务表，它有transaction_date，我想知道我的许多交易是如何在每个交易中发生的，不包括已取消的交易。

查询示例

SELECT transaction_id, transaction_status, 
(SELECT COUNT(*) FROM transaction WHERE user_id = 2 
AND transaction_date BETWEEN DATE_FORMAT(transaction_date, '%Y-%m-01 00:00:00') 
AND transaction_date AND transaction_status != 'cancel') AS transaction_counter
FROM transaction WHERE user_id = 2

Result:

-------------------------------------------------------------
| transaction_id | transaction_status | transaction_counter |
-------------------------------------------------------------
| 1              | success            | 1                   |
| 2              | cancel             | 1                   |
| 3              | success            | 2                   |
| 4              | success            | 3                   |
| 5              | success            | 4                   |
| 6              | cancel             | 4                   |
-------------------------------------------------------------

但是这个查询运行速度非常慢，因为它计算每一行，并且需要5-10分钟才能进行一千次交易。有没有想过优化这个查询？