I am baffled about the data window expected for both the training and test dataset.
The provided files for both training and test include the Longitude and Latitude. I think including the start and end dates in these two files will clarify all my doubts and help others who may be confused like myself.
This is important to ensure that we are indeed matching the expected time window during training with the provided training targets. It will also ensure that we are predicting the right time windows during testing.
Thanks for your question! Regarding Afghanistan, the time window to be considered is 1 to 30 April. Trainining and testing data refer to that period, and the model is expect to classify the cropland in the same period. For Iran and Sudan instead, the results should reflect the cropland over the full year.
What year? for both the training and test datasets?
This dataset is highly implicitly described. Unclear to some like myself and gives me (us) an undue disadvantage in making my (our) hands dirty.
Can the admin clarify it? I feel the data is poorly described and needs a better presentation for us all to participate actively.
Can someone who really understands this competition data statement help? @Lolletti, @amyflorida626, @Zindi
My present thought about the data window is as follows but not sure if this is right. Anyone with an exact time window should advise.
Training: 2019-07-01 ==> 2020-06-30Testing: 2022-07-01 ==> 2023-06-30Thanks for the query.
Data for Afghanistan, both for training and testing were collected between 1 and 30 April 2022.
Thanks .... then the right time range is
Training: 2019-07-01 ==> 2020-06-30Testing: 2022-07-01 ==> 2023-06-30Hey @Lolletti and @Zindi team,
Can we confirm @HungryLearner's conclusions to ensure we're on the same page? Understanding the dataset's date ranges is challenging, and given the nature of the data, it's time-consuming to acquire. It would be a time-saver for all of us.
Hey @Zindi, @amyflorida626,
There is a new discovery about the dataset from @Lolletti comments that "Data for Afghanistan, both for training and testing were collected between 1 and 30 April 2022." So we are now sure that for afghan, we are NOT GOING BEYOND 2022 for both training and testing and the month is strictly April and NO MAY.
Is it possible to now have the following phrases edited accordingly or else provide clarification to clear my misconception?
Please also help clarify this conclusion for sudan and iran:
Hi, thanks for your question. It was considered an issue uploading datasets for 2 years in Afghanistan within the same .csv, since it might have caused confusion. Therefore, only 2022 data were considered. The sentence in the evaluation and data sections will be amended accordingly.
Data for Afghanistan are ALL (both train and test) collected between 1 and 30 April 2022.
@Lolletti can you also clarify the dates (both train and test) for Sudan and Iran?
Hi, for Iran and Sudan please stick with what was explained in the info: July 2019 - June 2020.
Thanks @Lolletti for reply. so final clarification, does it mean for BOTH TRAIN AND TESTING for SUDAN and IRAN we stick to July 2019 - June 2020?
For Afghan, it is so clear now , thanks to you @Lolletti for clarification. For IRAN/SUDAN we are still stuck on the TRAIN/TEST period (s). Is is it same period -July 2019 - June 2020? You have'nt said clearly, but for afghan you have clearly repeated, so no doubt now.
@lolletti, I hope this one is fine now. Please comment on the two sections and not only on one. Thanks
Training & Testing: 2019-07-01 ==> 2020-06-30Is this what we should use @everyone here in this challenge
Yes... based on the last confirmation by host up to this point
Thank you