I would love to know how the rest of you are approaching the target variable, How you are coming up with the total number of turtles per capture site just prior to training.
In my case, I am using grouping by based on the expected results in the sample submission file. Grouping by year and week then counting the number the turtles saved. For the other variables, I am aggregating them based on what would seem viable. For example, mean for the weight variable. Counting categories of categorical variables.
Thank you so much, this has opened some lights for me
📷
📷
.
At first, I was grouping by capture site but I realized there are only 29 unique values so I was stuck there. Let me try your way and see how that turns out.
Hey Kipsang,
In my case, I am using grouping by based on the expected results in the sample submission file. Grouping by year and week then counting the number the turtles saved. For the other variables, I am aggregating them based on what would seem viable. For example, mean for the weight variable. Counting categories of categorical variables.
I hope this helps to shed some light.
Thank you so much, this has opened some lights for me
At first, I was grouping by capture site but I realized there are only 29 unique values so I was stuck there. Let me try your way and see how that turns out.
Glad to help!