Inundata: Mapping Floods in South Africa
'%3e%3cpath fill='%23001489' d='M0 0v6h9V0z'/%3e%3cpath fill='%23e03c31' d='M0 0v3h9V0z'/%3e%3cg stroke='%23fff' stroke-width='2'%3e%3cpath id='d' d='m0 0 4.5 3L0 6m4.5-3H9'/%3e%3cuse xlink:href='%23b' stroke='%23ffb81c' clip-path='url(%23c)'/%3e%3c/g%3e%3cuse xlink:href='%23d' fill='none' stroke='%23007749' stroke-width='1.2'/%3e%3c/g%3e%3c/svg%3e)
Helping South Africa
$10 000 USD
Completed (~1 year ago)
Classification
1340 joined
315 active
Start
Nov 22, 24
Close
Feb 16, 25
Reveal
Feb 17, 25
event_id as a feature
Help · 28 Jan 2025, 18:31 · 1
I originally thought that the same event_ids would be present in the train and test sets, however upon further inspection this does not seem to be the case. Is it safe to assume that our model is not expected to predict based on the location of observed precipitation? So we're only concerned with precipitation trends across all locations on the same day. I'm not too experienced with data science and AI/ML so I'm not sure how we could train a model based on one set of locations and then have it predict on a different set of unseen locations, if that makes sense. Thanks!
I believe the goal intended by the organizers is to have a model that, given a place with certain morphological characteristics, and a sequence of precipitation data, is able to generally assess if and when a flooding event will happen. The model learns from some location images with the relative precipitation time series (that could be registered at different times of the year and in different years, so not always in the same days) and predicts a flooding event for a new location, with new precipitation data. Quite ambitious task with the scarce data provided, but yeah. Hope I answered your question