Problem Statement: -

Find out cumulative purchases of each product over time.

Dataframe API Solution: -

# Define schema for the DataFrame 
schema = StructType([ StructField("order_id", IntegerType(), True), StructField("product_type", StringType(), True), StructField("quantity", IntegerType(), True), StructField("order_date", StringType(), True), ])
# Define data # Define data 
data = [ (213824, 'printer', 20, "2022-06-27"), (212312, 'hair dryer', 5, "2022-06-28 "), (132842, 'printer', 18, "2022-06-28"), (284730, 'standing lamp', 8, "2022-07-05") ]

df = spark.createDataFrame(data,schema)
df.show()

+--------+-------------+--------+-----------+
|order_id| product_type|quantity| order_date|
+--------+-------------+--------+-----------+
|  213824|      printer|      20|2022-06-27 |
|  212312|   hair dryer|       5|2022-06-28 |
|  132842|      printer|      18|2022-06-28 |
|  284730|standing lamp|       8|2022-07-05 |
+--------+-------------+--------+-----------+

window_spec = Window.partitionBy("Product_type").orderBy("order_date").rowsBetween(Window.unboundedPreceding,Window.currentRow)

result_df = df.withColumn("Cum_Purchased",sum("Quantity").over(window_spec)) \
.select("order_date","Product_type","Cum_Purchased")

result_df.show()

+-----------+-------------+-------------+
| order_date| Product_type|Cum_Purchased|
+-----------+-------------+-------------+
|2022-06-28 |   hair dryer|            5|
|2022-06-27 |      printer|           20|
|2022-06-28 |      printer|           38|
|2022-07-05 |standing lamp|            8|
+-----------+-------------+-------------+

Spark SQL Solution: -

df.createOrReplaceTempView("products")

result = spark.sql("""
SELECT
    Order_date,
    Product_type,
    SUM(Quantity) OVER (PARTITION BY Product_Type ORDER BY Order_date) AS Cum_Purchased
FROM products
""")
result.show()