reportforapriori

REPORT ON APPLYING THE APRIORI ALGORITHM

1.CREATING THE SPARSE MATRIX
  
  1.1 After creating the sparse matrix we noticed that the dataset had 18383 transactions (rows) and 37086 items (columns) in sparse format.
  
  1.2 Each and every row contained  the Invoice Number and Description. i.e the name of the products which the customer bought.
  
  1.3 example of a row is givven below
  
      {536366,                                                                                           HAND WARMER UNION JACK,HAND WARMER RED POLKA DOT}      

2.CHECKING FREQUENCY OF ITEMS BOUGHT  

  2.1 We checked the frequency of a particular item.
  
  2.2 For example frequency of  COLOUR SPACEBOY PEN is found to be  0.0001631943 
  
  2.3 Bar graphs were plotted to check the most frequently bought items
  
  2.4 From the graph it was found that BLACK RECORD COVER FRAME was the most frequentlly bought item followed by CHILLI LIGHTS,RABBIT NIGHT LIGHT,WHITE HANGING HEART T-LIGHT HOLDER...
  
  
3.THE APRIORI ALGORITHM

  3.1 On apllying the apriori algorithm with support value 0.0002 and confidence 60% we generated 638 rules.
  
  3.2 With Increase in suypport value to 0.0005 it generated only 4 rules
  
  3.3 so support was kept as 0.0002
  
  3.4 then best associations where found out 
  
  3.5 the top association was {6 RIBBONS ELEGANT CHRISTMAS,                                                           6 RIBBONS RUSTIC CHARM}      => {6 RIBBONS EMPIRE} with 0.0002175923 support,0.8 confidence and 3676.6 lift.
  
  3.6 similarly lots of association are studied
  
  3.7 Then we found that there are 391 redundant rules 
  
  3.8 On removing those redundant rules we obtained 247 rules.
  
  3.9 various graphs are plotted.
  
4.CONCLUSION
  
  4.1 On seeing the various association rules intersting combo offers can be given.