Previous

[🏠 Home

Next

User Manual

Key concepts in each link were briefly mentioned to save your valuable time. Click on the necessary links to know more.

1. About PAMI: Motivation and the people who supported

   PAMI is a PAttern MIning Python library to discover hidden patterns in Big Data.

2. Installation/Update/uninstall PAMI

     pip install pami
   

3. Organization of algorithms in PAMI

   The algorithms in PAMI are organized in a hierarchical structure as follows:

    PAMI.theoriticalModel.basic/maximal/closed/topk.algorithmName
   

4. Format to create different databases

   1. Transactional database

       format: item1<sep>item2<sep>...<sep>itemN
      

   2. Temporal database

       format: timestamp<sep>item1<sep>item2<sep>...<sep>itemN
      

   3. Neighborhood database

       format: spatialItem1<sep>spatialItem3<sep>spatialItem10<sep>...
      

   4. Utility database

       format: item1<sep>...<sep>itemN:totalUtility:utilityItem1<sep>...<sep>utilityItemN
      

   Default separator used in PAMI is tab space. However, users can override the separator with their choice.

5. Getting the statistics of databases

   1. Statistics of a transactional database This program outputs the statistical details of a transactional database. It will also output the distribution of items’ frequencies and transactional lengths.

      from dbStats import TransactionalDatabase as tds
      obj=tds.TransactionalDatabase(inputFile,sep)
      obj.run()
            
      #Getting basic stats of a database
      print("Database Size: " + obj.getDatabaseSize())
      print("Minimum transaction length:" + obj.getMinimumTransactionLength())
      print("Maximum transaction length:" + obj.getMaximumTransactionLength()) 
      print("Average transaction length:" + obj.getAverageTransactionLength())
      print("Standard deviation of transaction lengths:" + obj.getStandardDeviationTransactionLength())
            
      #Distribution of items' frequencies and transactional lengths
      itemFrequencies = obj.getItemFrequencies() #format: <item: freq>
      tranLenDistribution = obj.getTransanctionalLengthDistribution()  #format: <tranLength: freq>
      obj.storeInFile(itemFrequencies,outputFileName)
      obj.storeInFile(tranLenDistribution,outputFileName)        
      

   2. Statistics of a temporal database This program outputs the statistical details of a temporal database. It will also output the distribution of items’ frequencies, transactional lengths, and number of transactions occurring at each timestamp.

      from dbStats import TemporalDatabase as tds
      obj=tds.TemporalDatabase(inputFile,sep)
      obj.run()
            
      #Getting basic stats of a database
      print("Database Size: " + obj.getDatabaseSize())
      print("Minimum transaction length:" + obj.getMinimumTransactionLength())
      print("Maximum transaction length:" + obj.getMaximumTransactionLength()) 
      print("Average transaction length:" + obj.getAverageTransactionLength())
      print("Standard deviation of transaction lengths:" + obj.getStandardDeviationTransactionLength())
      print("Minimum period:" + obj.getMinimumPeriod())
      print("Maximum period:" + obj.getMaximumPeriod())
      print("Average period:" + obj.getAveragePeriod())
           
      #Distribution of items' frequencies, transactional lengths, and distribution of transactions per timestamp
      itemFrequencies = obj.getItemFrequencies() #format: <item: freq>
      tranLenDistribution = obj.getTransanctionalLengthDistribution()  #format: <tranLength: freq>
      transactionsPerTimestamp = obj.getNumberOfTransactionsPerTimestamp() #format: <timestamp: freq>
      obj.storeInFile(itemFrequencies,outputFileName)
      obj.storeInFile(tranLenDistribution,outputFileName) 
      obj.storeInFile(transactionsPerTimeStamp,outputFileName)
      

6. Converting dataframes to databases

   1. Format of dense dataframe

      tid/timestamp<sep>item1<sep>item2<sep>...<sep>itemN
      

   2. Format of sparse dataframe

      tid/timestamp<sep>item<sep>value
      

   3. Dataframe to database conversion

      This program creates a database by specifying a single condition and a threshold value for all items in a database. Code to convert a dataframe into a transactional database:

      from PAMI.DF2DB import DF2DB as pro
      db = pro.DF2DB(inputDataFrame, thresholdValue, condition, DFtype)
      # DFtype='sparse'  or 'dense'. Default type of an input dataframe is sparse
      db.createTransactional(outputFile)
      

   4. Dataframe to database conversed advanced

      This program user can specify a different condition and a threshold value for each item in the dataframe. Code to convert a dataframe into a transactional database:

      from PAMI.DF2DB import DF2DBPlus as pro
      db = pro.DF2DBPlus(inputDataFrame, itemConditionValueDataFrame, DFtype)
      # DFtype='sparse'  or 'dense'. Default type of an input dataframe is sparse
      db.createTransactional(outputFile)
      

   5. Spatiotemporal dataframe to databases

7. Exceuting Algorithms in PAMI

   1. Importing PAMI algorithms into your program

      from PAMI.frequentPattern.basic import fpGrowth  as alg
      obj = alg.fpGrowth(inputFile,minSup,sep)
      obj.startMine()
      obj.save('patterns.txt')
      df = obj.getPatternsAsDataFrame()
      print('Runtime: ' + str(obj.getRuntime()))
      print('Memory: ' + str(obj.getMemoryRSS()))
      

   2. Executing PAMI algorithms directly on the terminal

      1. Download the PAMI-main.zip file from GitHub
      2. Unzip the PAMI-main.zip file.
      3. Enter into the PAMI-main folder and move the PAMI sub-folder to the location of your choice.
      4. Let the location be /home/username/PAMI
      5. Execute the following command:

      python PAMI/patternModel/patternType/algorithm.py inputFile outputFile parameters
            
      E.g., python PAMI/frequentPattern/basic/fpGrowth.py inputFile outputFile minSup