FSPGrowth

class PAMI.georeferencedFrequentPattern.basic.FSPGrowth.FSPGrowth(iFile, nFile, minSup, sep='\t')[source]

Bases: _spatialFrequentPatterns

Description:

Given a transactional database and a spatial (or neighbourhood) file, FSPM aims to discover all of those patterns that satisfy the user-specified minimum support (minSup) and neighbourhood (maxDist) constraints. A pattern is spatially valid when all of its items are mutual neighbours.

This implementation preserves the public API of the original PAMI FSPGrowth class but replaces the earlier tree/prefix internals with a compact, parent-linked FP-tree, header-table mining, weighted conditional pattern bases, dictionary-based frequency ordering, and recursive spatial pruning that enforces the neighbourhood constraint during recursion rather than only at output time. It produces the same set of patterns and supports as the original.

Reference:

Rage, Uday & Fournier Viger, Philippe & Zettsu, Koji & Toyoda, Masashi & Kitsuregawa, Masaru. (2020). Discovering Frequent Spatial Patterns in Very Large Spatiotemporal Databases.

Parameters:
  • iFile – str or pd.DataFrame : Name/path of the input transactional database, or a DataFrame containing a ‘Transactions’ column.

  • nFile – str or pd.DataFrame : Name/path of the input neighbourhood file, or a DataFrame containing item and neighbour columns.

  • minSup – int or float or str : The user can specify minSup either as a count or as a proportion of the database size. If the data type of minSup is integer, it is treated as a count. Otherwise (float or decimal string), it is treated as a proportion of the number of transactions.

  • sep – str : Separator used to distinguish items within a transaction. Default is a tab space; users may override.

Attributes:
iFilestr or pd.DataFrame

Input file name/path of the transactional database.

nFilestr or pd.DataFrame

Input file name/path of the neighbourhood file.

oFilestr

Name/path of the output file.

minSupint or float

Minimum support, as a count or as a proportion of database size.

finalPatternsdict

Complete set of discovered patterns: tab-joined pattern string -> support.

startTimefloat

Start time of the mining process.

endTimefloat

Completion time of the mining process.

memoryUSSfloat

Total USS memory consumed by the mining process.

memoryRSSfloat

Total RSS memory consumed by the mining process.

Methods:
mine()

Starts the pattern-mining process.

getPatterns()

Returns the complete set of patterns as a dictionary.

save(oFile)

Writes the complete set of patterns to an output file.

getPatternsAsDataFrame()

Returns the complete set of patterns as a pandas DataFrame.

getMemoryUSS()

Returns the total USS memory consumed by the mining process.

getMemoryRSS()

Returns the total RSS memory consumed by the mining process.

getRuntime()

Returns the total runtime taken by the mining process.

printResults()

Prints a summary of the results.

Executing the code on terminal :

Format:

(.venv) $ python3 FSPGrowth.py <inputFile> <outputFile> <neighbourFile> <minSup>

Example Usage:

(.venv) $ python3 FSPGrowth.py sampleTDB.txt output.txt sampleN.txt 0.5

Note

A float minSup is interpreted as a proportion of the database transactions.

Sample run of importing the code :

from PAMI.georeferencedFrequentPattern.basic import FSPGrowth as alg

obj = alg.FSPGrowth("sampleTDB.txt", "sampleN.txt", 5)

obj.mine()

spatialFrequentPatterns = obj.getPatterns()

print("Total number of Spatial Frequent Patterns:", len(spatialFrequentPatterns))

obj.save("outFile")

memUSS = obj.getMemoryUSS()

print("Total Memory in USS:", memUSS)

memRSS = obj.getMemoryRSS()

print("Total Memory in RSS", memRSS)

run = obj.getRuntime()

print("Total ExecutionTime in seconds:", run)

Credits:

The original program was written by Yudai Masu under the supervision of Professor Rage Uday Kiran. This optimized, API-compatible implementation retains that public interface while replacing the internal mining engine.

getMemoryRSS()[source]

Total amount of RSS memory consumed by the mining process.

Returns:

RSS memory consumed.

Return type:

float

getMemoryUSS()[source]

Total amount of USS memory consumed by the mining process.

Returns:

USS memory consumed.

Return type:

float

getPatterns()[source]

Return the complete set of frequent patterns after mining.

Returns:

Patterns as { tab-joined pattern string : support }.

Return type:

dict

getPatternsAsDataFrame()[source]

Store the discovered frequent spatial patterns in a DataFrame.

Returns:

Patterns as a DataFrame with columns [‘Patterns’, ‘Support’].

Return type:

pd.DataFrame

getRuntime()[source]

Total runtime taken by the mining process.

Returns:

Runtime in seconds.

Return type:

float

mine()[source]

Start the pattern-mining process.

printResults()[source]

Print a summary of the results.

save(oFile)[source]

Write the complete set of frequent patterns to an output file.

Parameters:

oFile (str) – Name of the output file.

startMine()[source]

Start the pattern-mining process (deprecated alias for mine()).

class PAMI.georeferencedFrequentPattern.basic.FSPGrowth.SpatialFPGrowth(iFile, nFile, minSup, sep='\t')[source]

Bases: FSPGrowth