How does MapReduce work, specifically the Reduce stage?

MapReduce is a programming model for processing large data sets, consisting of two main stages: Map and Reduce.

In the Map stage, input data is split into key-value pairs, which are processed in parallel. Each Map function transforms the input data into intermediate key-value pairs.

The Reduce stage collects all intermediate values with the same key and aggregates them, performing a summary operation (e.g., summing, counting, merging).

Example: counting the number of words in a large text.

• Map: emit (word, 1) for each word
• Shuffle: grouping by word
• Reduce: summing all units for each word

Thus, Reduce aggregates data received from Map and produces the final result.