none
Problem using Microsoft Association Rules with a single table

    Question

  • I have a table with 3 columns.

    ITEM_ID ITEMNUM    ITEM
    1           1001           Chocolates
    2           1001           Pencil
    3           1001          paper
    4           1001          Marker
    5           1001          plums
    6           1002          Pencil

    and 44 more rows

    I am trying to create the data mining structure for the Microsoft Association Rules using the data mining wizard.

    The wizard asks for a table, I only have the table above.

    The wizard asks to specify the training data and gives three choices, key, input predictable.

    I chose ITEMNUM as key and chose ITEM as both input and predictable. This is market basket analysis and I am trying to find with items occur with each other.

    I am getting this error.

    The Association Rules algorithm found no rules matching the algorithm parameters for model M Basket2.

    What am I doing wrong


    • Edited by Al276276 Tuesday, June 25, 2013 4:38 PM
    Tuesday, June 25, 2013 6:32 AM

Answers

All replies

  • Try to reduce values of MINIMUM_PROBABILITY and MINIMUM_SUPPORT parameters for the association rules algorithm.

    Tatyana Yakushev [PredixionSoftware.com]

    Download Predixion Insight 3.0 - World class predictive platform for big data

    Tuesday, June 25, 2013 4:15 PM
    Answerer
  • Recommend you Edit your Thread title so it is not marked as Abusive and deleted.

    Chris Ward

    Tuesday, June 25, 2013 4:18 PM
  • Tatyana,

    Thank you for responding.

    I am trying to understand how to create the data mining structure for Microsoft Association Rules technique using Visual Studio Business Intelligence Development Studio 2008. I have a few books on data mining and have looked on the internet, but I cannot seem to find a simple example of how to use a single table with 2 main columns to produce Microsoft Association Rules results using the business intelligence wizards in Visual Studio 2008.

    I have included the simple table that I mentioned in my first post in the event someone would like to try the same example. (ITEM_ID is just a primary key and ignored, ITEMNUM is the basket number, ITEM is the actual item purchased).

    My first step was to test the table in Excel using “Microsoft® SQL Server® 2012 Data Mining Add-ins for Microsoft® Office® 2010”  to make sure that the table structure was valid.  When I run that table in Excel 2010 using the “Associate Model” using:

    "Minimum Support" of 2,  

    "Minimum rule probability" of 20.0,

    “Transaction ID” using ITEMNUM and  

    “Item” using ITEM

    I get the following EXCEL partial Data Mining results, which suggests to me that the short table is capable of working in the VS 2008 data mining models.

    Probability

    Importance

    Rule

    100 %

    0.35

    Tomatoes = Existing, paper = Existing -> Marker = Existing

    100 %

    0.53

    Tomatoes = Existing -> Oranges = Existing

    100 %

    0.13

    Tomatoes = Existing, Marker = Existing -> paper = Existing

    100 %

    0.35

    Tomatoes = Existing -> Marker = Existing

    100 %

    0.13

    Tomatoes = Existing, Oranges = Existing -> Choclates =   Existing

    100 %

    0.13

    Tomatoes = Existing -> paper = Existing

    100 %

    0.05

    Tomatoes = Existing, Choclates = Existing -> Pencil =   Existing

    Now, when I use the Data Mining Wizard in VS 2008,

    1. it asks me for the table. I check the “case” box and leave the “nested” box blank
    2. The wizard then presents the three columns in my table (ITEM_ID, ITEMNUM, ITEM). I ignore the ITEM_ID since it is only a row placeholder.
    3. I select ITEMNUM as the “key”
    4. I select ITEM as both “input” and “predictable”  ( I has not found a good definition for these terms in my books or on the internet, but the Wizard forces me to select both columns before it will allow me to continue)
    5. I use “Percentage of Data for Testing” 50%
    6. MINIMUM _SUPPORT is has a default of 0
    7. MINIMUM_PROBABILITY has a default of .40, but taking your suggestion I set it to .20 , the same as I used in the EXCEL model
    8. I have tried the Data Mining wizard using many other values ( MINIMUM _SUPPORT = 0 to 2 and MINIMUM_PROBABILITY = .0001 to .30.
    9. and I have doubled the size of the table to 88 rows, but the results are always the same. “The Association Rules algorithm found no rules matching the algorithm parameters for model”

    10.  If anyone has any links to explain how to run this model or can explain what “input” and “predictable” mean in a two-column table I would be appreciative.

    Thanks

    Al

    ITEM_ID

    ITEMNUM

    ITEM

    1

    1001

    Choclates

    2

    1001

    Pencil

    3

    1001

    paper

    4

    1001

    Marker

    5

    1001

    plums

    6

    1002

    Pencil

    7

    1002

    Choclates

    8

    1003

    Pencil

    9

    1003

    Coke

    10

    1003

    Eraser

    11

    1004

    Pencil

    12

    1004

    Choclates

    13

    1004

    Paper

    14

    1004

    Cookies

    15

    1005

    Marker

    16

    1006

    Pencil

    17

    1006

    Marker

    18

    1007

    Pencil

    19

    1007

    Choclates

    20

    1007

    paper

    21

    1008

    Choclates

    22

    1008

    Pencil

    23

    1008

    paper

    24

    1008

    plums

    25

    1008

    Cookies

    26

    1009

    Cookies

    27

    1009

    Oranges

    28

    1009

    Tomatoes

    29

    1010

    Oranges

    30

    1010

    Cookies

    31

    1011

    Oranges

    32

    1011

    Tomatoes

    33

    1011

    Choclates

    34

    1011

    Pencil

    35

    1011

    paper

    36

    1011

    Marker

    37

    1011

    plums

    38

    1012

    Tomatoes

    39

    1012

    Choclates

    40

    1012

    Pencil

    41

    1012

    paper

    42

    1012

    Marker

    43

    1012

    Oranges

    44

    1012

    Tuna


    AAC

    Wednesday, June 26, 2013 4:34 AM
  • Thanks Chris for pointing out my title error. I forgot to fix the title before submitting the question and have corrected it.

    Al


    AAC


    • Edited by Al276276 Wednesday, June 26, 2013 4:36 AM added to the sentence
    Wednesday, June 26, 2013 4:35 AM
  • Sunday, November 17, 2013 12:19 PM