• PCY algorithm exploits the observation that there may be much unused space in main memory on the first pass of apraisy.

• In first pass only a bash function is applied on pair of item so that they bashes to a bucket.

• we hash each pair and add 1 to the bucket.

• At the end of 1st pass each bucket has count.

• If count of bucket is > = support then the bucket is frequent bucket.

• we can define a candidate pair C2 to be those pair { i , j } such that

1) i and j are frequent items.

2) { i , j } hashes to a frequent bucket.

• Multistage uses several successive hash tables to reduce further the number of candidate pairs.

• But requires more passes than normal PCY.