Question

是否有一种有效的方法可以对频率大约为13-15分钟到15分钟的数据进行上采样/重采样。我有多个id和200M +行。

dataframe=spark.createDataFrame([("J1", "2019-12-29 12:07:38", 100), ("J1", "2019-12-29 12:24:25", 200), 
                          ("J1", "2019-12-29 12:37:58", 100), ("J8", "2020-09-09 13:06:36", 300), 
                          ("J8", "2020-09-09 13:21:37", 200), ("J8", "2020-09-09 13:36:38", 400)], 
                          ["id", "date_time", "some_value"]).show()

+---+-------------------+----------+
| id|               date|some_value|
+---+-------------------+----------+
| J1|2019-12-29 12:07:38|       100|
| J1|2019-12-29 12:24:25|       200|
| J1|2019-12-29 12:37:58|       100|
| J8|2020-09-09 13:06:36|       300|
| J8|2020-09-09 13:21:37|       200|
| J8|2020-09-09 13:36:38|       400|
+---+-------------------+----------+

所需数据框：

+---+-------------------+----------+
| id|               date|some_value|
+---+-------------------+----------+
| J1|2019-12-29 12:15:00|       100|
| J1|2019-12-29 12:30:00|       200|
| J1|2019-12-29 12:45:00|       100|
| J8|2020-09-09 13:00:00|       300|
| J8|2020-09-09 13:15:00|       200|
| J8|2020-09-09 13:30:00|       400|
+---+-------------------+----------+

Answer 1

有一个功能window。它同时生成start和end。您可能需要应用其他功能来选择最接近的。

from pyspark.sql import functions as F

df.withColumn("date_time", F.window("date_time", "15 minutes")["end"]).show()
+---+-------------------+----------+
| id|          date_time|some_value|
+---+-------------------+----------+
| J1|2019-12-29 12:15:00|       100|
| J1|2019-12-29 12:30:00|       200|
| J1|2019-12-29 12:45:00|       100|
| J8|2020-09-09 13:15:00|       300|
| J8|2020-09-09 13:30:00|       200|
| J8|2020-09-09 13:45:00|       400|
+---+-------------------+----------+

PySpark-上采样/重采样时间序列数据

1 个答案: