Recognising Patterns

Map clues to techniques

Most interview problems reuse a small set of patterns. Clues in the wording point to them: "sorted array" suggests binary search or two pointers, "contiguous subarray" suggests sliding window or prefix sums, "top k" suggests a heap, "all combinations" suggests backtracking, "minimum steps in a grid" suggests BFS, "prerequisites" suggests topological sort, and "number of ways / minimum cost" suggests dynamic programming.

Clue to pattern cheat sheet

A starting point, not a rule.

pair / complement / seen before        -> hash map
subarray sum equals k                  -> prefix sums + hash map
sorted input, pair or triplet          -> two pointers
longest/shortest contiguous window     -> sliding window
sorted / monotonic answer space        -> binary search
next greater / smaller element         -> monotonic stack
cycle or middle of linked list         -> fast & slow pointers
shortest path, unweighted              -> BFS
connected components                   -> DFS / union-find
dependencies / ordering                -> topological sort
weighted shortest path                 -> Dijkstra
top k / k-th largest / merge k lists   -> heap
overlapping ranges                     -> sort + intervals
all subsets / permutations             -> backtracking
count ways / min cost / max value      -> dynamic programming

Practise by pattern, then mix

Solve several problems per pattern to learn it, then practise mixed sets so you learn to recognise patterns cold.

Quick check: Which pattern does "longest substring with at most k distinct characters" suggest?

• Backtracking
• Topological sort
• Dijkstra
• Sliding window