PAKDD 2007 Data Mining Competition
PAKDD competition

  Overview

  Participation Guidelines

  Evaluation and Awards

  Dates and Updates

  Results and Reports

  Final Winners

  User Login

  Downloads

  Online Registration

  Online Submission

  PAKDD 2007

  Contact

Important Dates

From 15 February 2007 -Dataset available, please see the participation guidelines page for online registration information.

15 April 2007 (11:59PM +8GMT China Standard Time) -Deadline for competition entries to be submitted online.

1 May 2007 -Competition results to be published on this website, and winners to be notified.

23 May 2007 -Awards to be presented to the winning teams at the PAKDD 2007 conference.

Spring 2008 -Selected extended papers will be published in the International Journal of Data Warehousing and Mining

 

Competition Updates for Participants

13) If you do not remember your participant id, then please contact our webmaster and inform us of the email address of the main registrant and we will send the participant id together with a new password to that email address.

12) If you forget your password, then please contact our webmaster and inform us of the participant id, and we will reset the password and send a new one to the main registrant's email address.

11) As we might possibly be evaluating the reports later in case there might be ties, we would like all reports to be in a comparable format. It will complicate the judging if some reports come in document format, and some in slides format, as the two writing styles can be very different. Also to clarify, when we say "PDF format", we mean PDF rendering of a word processor document format technical report, not PDF of a presentation slides style report. If you don't have Word, then we suggest you just use whatever word processing software you have, and then convert to PDF using some free PDF converter downloadable from the Internet.

10)If you would like to replace a submission with a newer one, or if you think the previous submission might have been corrupted or incomplete, then please contact our webmaster to clear your submission record (please specify if report and/or predictions file) so that you can submit again.

9) Wed, 21st March 2007 - Note that for bureau variables, the special values "Went to bureau and no match found (new file created)" and "Did not go to bureau" basically mean that there was no existing credit record with the credit bureau for this account holder. Participants can choose to treat these special values as a kind of missing value.

8) Tue, 20th March 2007 - Note that "Bureau" in Data Dictionary Ver4 refers to Credit Bureau, which is an external agency that coordinates the sharing of information between companies regarding which customers have good or bad credit histories and status.

7) Wed, 7th March 2007 - Correction to Data Dictionary Ver4: Note that for A_DISTRICT_APPLICANT, valid values are from '0' to '8' (not '1' to '9').

6) Please note that all the variables provided are as of date of application for the company's credit card product. So they would be not inclusive of that credit card product itself (but which we know they will be approved for later for them to be in the datasets).

5) Mon, 26th Feb. 2007 - Paragraph 2 of Evaluation and Awards Section has been updated with a sample template of how to approximate the Area Under the ROC Curve, as well as some suitable web reference regarding the metric.

4) Sun, 25th Feb. 2007 -- Please note there were minor errors in 2 of the definitions from Data Dictionary VER3.

(4.a)TARGET_FLAG supposed to be "TARGET FIELD, indicating if the credit card customer, who did not have an existing home loan with the company before when account opened, signed up for a new home loan with the company within 12 months of opening the credit card account."

(4.b)Also, the correct definition for AGE_AT_APPLICATION is supposed to be "Indicates customer's age (in years) upon application for company's credit card product."

The updated archive file, "Competition_Downloads_VER4" in the Downloads section now contains Data_Dictionary_VER4 reflecting the two changes above.

3) Submissions section paragraph (1) corrected to state that "8,000 scores" are required.

2) Datasets and Data Dictionary will only be available from 15 Feb 2007.

1) Wed, 7th Feb, 2007 - Evaluation and Awards page has been modified. Please check to see latest instructions.

  

Copyright @2007 LeVis Group, Shanghai University
Webmaster : Li Dan - Shanghai University (China)