A daily silver job joins about 1 million sales rows to dim_prices, which holds one row per (product_id, store_id) because prices vary by store. After the join runs, the silver table contains about 3 million rows and revenue totals are inflated. The suspicious line: ```python silver = (sales .join(dim_prices, "product_id") .withColumn("revenue", col("qty") * col("price"))) ``` What is the most likely cause, and the correct fix?

1. A. The sales fact contains duplicate transactions — call dropDuplicates() with no subset before the join 💬 A
2. B. Spark picked the wrong physical join strategy — force a broadcast join with .hint('broadcast') so each sale matches exactly one price 💬 B
3. C. The join key is not unique in dim_prices, so sales fan out one row per store — join on both product_id and store_id ✓ C
4. D. dim_prices contains NULL prices that replicate rows during the join — filter out NULL prices before joining 💬 D

Reveal answer

Correct answer: The join key is not unique in dim_prices, so sales fan out one row per store — join on both product_id and store_id

Reference: https://docs.databricks.com/aws/en/pyspark/reference/functions