Build query 'average amount by month' with filling gaps from previous month - mysql

Question

I have a problem. I need to build report query to calculate average amount by month.
If user from account does not buy anything in this month I should use amount from previous month.
This is my data: As you can see I have one account operation on July (2014-07) and one operation in August (2014-08).
CREATE TABLE IF NOT EXISTS `value_hist` (
`id` int(11) NOT NULL,
`account_id` int(11) DEFAULT NULL,
`data_channel_id` int(11) DEFAULT NULL,
`status` tinyint(1) NOT NULL,
`amount` decimal(19,4) NOT NULL COMMENT '(DC2Type:money)(DC2Type:money)',
`created_at` datetime NOT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci AUTO_INCREMENT=8 ;
INSERT INTO `value_hist` (`id`, `account_id`, `data_channel_id`, `status`, `amount`, `created_at`)
VALUES
(28, 19, 2, 0, 100.0000, '2014-08-22 14:59:47'),
(29, 19, 2, 0, 11.0000, '2014-09-22 15:05:36'),
(30, 19, 2, 0, 22.0000, '2014-09-22 15:10:59'),
(31, 19, 2, 0, 0.0000, '2014-09-22 15:14:19'),
(32, 19, 2, 0, 22.0000, '2014-09-22 15:15:58'),
(33, 19, 2, 0, 0.0000, '2014-09-22 15:17:29'),
(34, 1, 2, 0, 50.0000, '2014-07-22 15:20:10'),
(35, 1, 2, 0, 0.0000, '2014-09-22 15:23:37'),
(36, 1, 2, 0, 100.0000, '2014-09-22 15:23:39'),
(37, 1, 2, 1, 0.0000, '2014-09-22 15:30:53'),
(38, 19, 2, 0, 100.0000, '2014-09-22 15:30:53'),
(39, 28, 2, 1, 0.0000, '2014-09-22 15:36:44'),
(40, 19, 2, 1, 0.0000, '2014-09-22 15:45:24'),
(41, 6, 2, 0, 0.0000, '2014-09-22 15:48:54'),
(42, 6, 2, 0, 200.0000, '2014-09-22 15:49:34'),
(43, 6, 2, 1, 33.0000, '2014-09-22 15:50:18');
I have built a query. Which can get average value from each month
SELECT
data_channel_id,
test2.monthaa,
AVG(h1.amount) as amount
FROM
value_hist h1
JOIN (
SELECT MAX(dd.`id`) as id, EXTRACT(month from dd.created_at) as monthaa
FROM value_hist dd
where dd.created_at > '01.01.2014'
GROUP BY data_channel_id, dd.account_id, monthaa
) test2 ON test2.id = h1.id
WHERE
created_at > '01.01.2014'
GROUP BY data_channel_id, monthaa
The result return average value from each month
| channel | month | amount |
| 2 | 7 | 50.0000|
| 2 | 8 | 100.000|
| 2 | 9 | 8.25000|
But I should get this result, I should consider previous month amount. For example (100 + 50) / 2, because account 1 did not buy in August (2014-08)
| channel | month | amount |
| 2 | 7 | 50.0000|
| 2 | 8 | 75.0000|
| 2 | 9 | 8.25000|
Update: There was only one account in July(7) thus amout equal to 50.00; (50.00 / 1 = 50.00).
In August(8) second account has come therefore amount equal to 75.00; ((100.00 + 50) / 2 = 75.00) because I have two accounts in August(8).