 | hlcm-0.1: Fast algorithm for mining closed frequent itemsets | Contents | Index |
|
|
|
| Description |
Library for using the LCM algorithm in order to compute closed frequent pattern.
Input must be a transaction database, either in text format (as a ByteString)
or in [[Item]] format, where Item = Int.
Several bencharking functions allowing to tune parallel strategy used and depth
cutoff are also provided.
|
|
| Synopsis |
|
|
|
| Documentation |
|
| type Frequency = Int |
|
| type Item = Int |
|
| runLCMstring |
| :: ByteString | The transaction database as a big string. Transactions are separated by newlines, items are separated by spaces
| | -> Frequency | Minimum frequency threshold for the frequent itemsets
| | -> [[Item]] | Output: list of closed frequent itemsets
| | Get the data as a long bytestring, parses it and
and executes LCM to discover closed frequent itemsets.
|
|
|
| runLCMmatrix |
| :: [[Item]] | The transaction database as matrix of items (List of List)
| | -> Frequency | Minimum frequency threshold for the frequent itemsets
| | -> [[Item]] | Output: list of closed frequent itemsets
| | Get the data as a matrix of Items, parses it and
and executes LCM to discover closed frequent itemsets.
|
|
|
| benchLCM_parBuffer |
| :: ByteString | The transaction database as a big string. Transactions are separated by newlines, items are separated by spaces
| | -> Frequency | Minimum frequency threshold for the frequent itemsets
| | -> Int | value for parBuffer
| | -> Int | depth for cutting parallelism
| | -> [[Item]] | Output: list of closed frequent itemsets
| Use for benchmarking, parallel strategy = parBuffer by Simon Marlow.
This strategy does not have space leak.
/Warning: outputs are unusable as is, because items are renamed internally, and in this function the reverse
renaming is not performed. It is trivial to have it back by copying the code from runLCMstring./
|
|
|
| benchLCM_parMap |
| :: ByteString | The transaction database as a big string. Transactions are separated by newlines, items are separated by spaces
| | -> Frequency | Minimum frequency threshold for the frequent itemsets
| | -> Int | depth for cutting parallelism
| | -> [[Item]] | Output: list of closed frequent itemsets
| Use for benchmarking, parallel strategy = parMap from Control.Parallel.Strategies.
/Warning: outputs are unusable as is, because items are renamed internally, and in this function the reverse
renaming is not performed. It is trivial to have it back by copying the code from runLCMstring./
|
|
|
| Produced by Haddock version 2.6.1 |