Pyspark subselect / subquery join using dataframes

Question

I am looking to join to a value based on the closest match below that value. In SQL I can do this quite easily. Consider the following data:

tblActuals

|Date       |Temperature:
|09/02/2020 |14.1
|10/02/2020 |15.3
|11/02/2020 |12.2
|12/02/2020 |12.4
|13/02/2020 |12.5
|14/02/2020 |11
|15/02/2020 |14.6

tblCoefficients:

|Metric |Coefficient
|10.5   |0.997825593
|11     |0.997825593
|11.5   |0.997663198
|12     |0.997307614
|12.5   |0.996848773
|13     |0.996468537
|13.5   |0.99638519
|14     |0.996726301
|14.5   |0.997435894
|15     |0.998311153
|15.5   |0.999135509

In SQL I can achieve joining this by using the below:

Select 
    a.date, 
    b.temperature, 
    (select top 1 b.Coefficient from tblCoefficients b where b.Metric <= a.Temperature order by b.Metric desc) as coefficient 
from tblActuals

Is there any way to achieve the same as the above with the data in two pyspark dataframes? I can achieve a similar result in spark SQL but I need the flexibility of the dataframes for the process I am creating in databricks.

Answer 1

You can do a join and get the coefficient for the maximum (closest) metric:

import pyspark.sql.functions as F

result = tblActuals.join(
    tblCoefficients,
    tblActuals['Temperature'] >= tblCoefficients['Metric']
).groupBy(tblActuals.columns).agg(
    F.max(F.struct('Metric', 'Coefficient'))['Coefficient'].alias('coefficient')
)

result.show()
+----------+-----------+-----------+
|      Date|Temperature|coefficient|
+----------+-----------+-----------+
|15/02/2020|       14.6|0.997435894|
|12/02/2020|       12.4|0.997307614|
|14/02/2020|       11.0|0.997825593|
|13/02/2020|       12.5|0.996848773|
|11/02/2020|       12.2|0.997307614|
|10/02/2020|       15.3|0.998311153|
|09/02/2020|       14.1|0.996726301|
+----------+-----------+-----------+