# reduce (10-reduce.com)

To make something smaller or simpler. People say this when they want less of something, like reducing stress, reducing clutter, or reducing costs.

## What a model may hear
- array reduction (functional programming (JavaScript, Python, etc.)): fold a list into a single value by applying a function cumulatively to each element
- MapReduce operation (distributed computing (Hadoop, Spark)): aggregate data across many machines after mapping, often summing or grouping results
- dimensionality reduction (machine learning and statistics): compress features or variables while preserving structure, as in PCA or t-SNE
- reduction in formal logic (mathematics and type theory): simplify an expression to a normal form through rewrite rules

## Where people and models part ways
- Says: "Reduce this for me"
  Means: Make this shorter or simpler to read
  May be taken as: Apply a mathematical reduction or data aggregation that changes the content unexpectedly
  Say instead: "Summarize this and make it shorter"
- Says: "Can you reduce the noise?"
  Means: Remove distracting parts from audio or text
  May be taken as: Apply a signal processing filter or statistical noise reduction that alters the source
  Say instead: "Remove the distracting background sounds from this audio"
- Says: "Reduce the list to what matters"
  Means: Pick the important items
  May be taken as: Run a fold operation that combines items mathematically rather than selecting
  Say instead: "Pick the most important items from this list"
- Says: "I need to reduce my data"
  Means: Make the dataset smaller to work with
  May be taken as: Apply PCA or feature selection automatically, changing what the data represents
  Say instead: "Make this dataset smaller by removing duplicate rows"

## Tips
- Say 'summarize' instead of 'reduce' when you mean make text shorter
- Say 'combine these into one number' if you want a sum or total
- Say 'pick the top items' if you want selection, not aggregation
- Specify 'remove noise from audio' rather than 'reduce noise' for sound editing
- Use 'compress the features' only if you want a machine learning transformation

## Often confused with
- summarize: produces a shorter version, not a single computed value
- compress: preserves all information for later recovery
- filter: removes some items, keeps others unchanged
- aggregate: combines many into one, often by summing or averaging
- simplify: makes easier to understand, not necessarily shorter
- fold: the functional programming synonym for reduce