PAKDD 2009 Data Mining Competition
Dates and Updates

Dates

Dates
Activities
Apr 27 Conference starts
Apr 03 Competition results released
Mar 20 Competition submission deadline (PDF manuscript and scores)
Mar 02 Prediction data set release
Prediction submission open (for manuscript and scores)
Feb 16 LeaderBoard data set release
LeaderBoard open for submissions
Feb 12 Competition page open
Modeling data set release
Competition announcement

Updates

Dates
Updates (newest on the top)
Jun 06 Today we have released new features in the competition website to support academic researching after competition closure. These new features include the registration of new accounts, the downloading of the competition data and new submissions for benchmarking purposes (no limit for submissions).
Apr 20 Availability of results for all competitors. Results visible only for team members.
Mar 19 Adjustment of the deadline to the Brazilian Standard Time (10 extra hours)
Link to the manuscript format
Forum for publication of selected and reviewed approaches
Mar 03 Due to demands from several competitors, we are expanding the limit of submission to the LeaderBoard to a maximum of 15 instead of the previous 10.
Feb 20 The Java code used to calculate the area under the ROC curve has been updated. This newer version uses trapezoids to compute the area (instead rectangles). This improvement allows more precise estimation.
Feb 19 Explanations about the data: Some competitors have raised questions about the data found in some variable fields and this update is just to explain that these data were collected from an actual relational database in operation. Only some fields identifying clients and region have been either removed or modified, as previously stated. The list below exemplifies some types of discrepancies found on the database along the 5 years of data collection.
* Some fields represent potentially useful information or future services that have never been implemented (e.g. FLAG_CARD_INSURANCE_OPTION).
* Other variables started being collected from a given moment onwards (e.g. QUANT_DEPENDANTS).
* Some other fields were not compulsory and applicants only filled them when demanded at the application desk, maybe according to a temporary policy.
* The PERSONAL_REFERENCE fields contain the first name (in Portuguese) of the personal references
Feb 16 LeaderBoard data set release
LeaderBoard open for submissions
Extra description on the variable list (Residence type)
AUC_ROC Java script fixed (posted for the teams which had already downloaded)
Feb 12 Competition page open
Modeling data set release
Competition announcement
Locations of visitors to this page