Azure Streaming Analytics计算OHLC

Question

我正在尝试使用Azure Stream Analytics SQL计算股票价格的开盘价，最低价，最高价和收盘价。

我可以很容易地得到Min和Max，但我很难弄清楚如何计算TumblingWindow的Open（FIRST）和Close（LAST）。

我已经在这里找到了文档（https://docs.microsoft.com/en-us/azure/stream-analytics/stream-analytics-stream-analytics-query-patterns），但我不能将它们组合成一个查询。

这就是我所拥有的：

SELECT
    DateAdd(second,-5,System.TimeStamp) as WinStartTime,
    system.TimeStamp as WinEndTime,
    exchange,
    Max(price) as H,
    Min(price) as L,
    Count(*) as EventCount
FROM [input1]
GROUP BY TumblingWindow(second, 5), exchange

我在SQL方面不是很先进，但有没有办法将它组合成一个查询？即使使用子查询。

注意：5秒只是我选择进行测试的一个示例窗口。

Answer 1

根据您的方案，我假设您可以利用Azure流分析中的Collect聚合函数和user-defined functions来实现您的目的。以下是详细信息，您可以参考它们：

假设您的输入如下：

[
 {"price":1.1,"exchange":10,"Time":"2017-7-24T13:00:00Z"},
 {"price":1.2,"exchange":20,"Time":"2017-7-24T13:04:00Z"},
 {"price":1.4,"exchange":40,"Time":"2017-7-24T13:03:00Z"},
 {"price":1.3,"exchange":30,"Time":"2017-7-24T13:02:00Z"},
 {"price":1.5,"exchange":50,"Time":"2017-7-24T13:06:00Z"}
] 

<强> UDF

// udf.getLast
function main(objs) {
    if(objs==null||objs==undefined||objs.length==0)return null;
    return objs[objs.length-1];
}

// udf.getFirst
function main(objs) {
    if(objs==null||objs==undefined||objs.length==0)return;
    return objs[0];
}

<强> QUERY

SELECT
    DateAdd(minute,-5,System.TIMESTAMP) as WinStartTime,
    System.TIMESTAMP as WinEndTime,
    UDF.getFirst(Collect()) AS FIRST,
    UDF.getLast(Collect()) AS LAST,
    Max(price) as H,
    Min(price) as L,
    Count(*) as EventCount
FROM [input1] TIMESTAMP By Time
GROUP BY TumblingWindow(minute, 5)

<强>结果：