I've noticed that there are several identical scores "112.4561452" on the list. I'm wondering if this is due to the data. I've been trying for a while now, but I'm still around 110 and haven't seen much improvement.
Maybe you are not considering the ID on train and test.
Separate the date from the id in the sample, and consider it by applying the same label encoder on train and test and train an LGBM you will easily reach 75.
A cautionary note...When using ID that appear in both the train and test (which is actually the ID in the sample submission file as there are no test set, unless of cause you split the train set and developed a test set), are u not guilty of committing a special form of 'LEAKAGE', in particular I believe it constitute a case of train-test contamination as data snooping occur through the use of ID variable, so if the same 'ID' that is in the train set is also in the test set(sample submission)...a cautionary note!!! So you will climb up the leader board only to be penalised later for leakage and overfitting....just sharing a thought!!!
'%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)
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e){kind=small}
112.45 is what you get if your submission is only zeros clicks per test sample
Focus, just like in real life, look out for shade(hidden anomolies and gremlins)!!!
Whenever I feel that I hit the wall, I start over from scratch.
Maybe you are not considering the ID on train and test.
Separate the date from the id in the sample, and consider it by applying the same label encoder on train and test and train an LGBM you will easily reach 75.
A cautionary note...When using ID that appear in both the train and test (which is actually the ID in the sample submission file as there are no test set, unless of cause you split the train set and developed a test set), are u not guilty of committing a special form of 'LEAKAGE', in particular I believe it constitute a case of train-test contamination as data snooping occur through the use of ID variable, so if the same 'ID' that is in the train set is also in the test set(sample submission)...a cautionary note!!! So you will climb up the leader board only to be penalised later for leakage and overfitting....just sharing a thought!!!