Zimnat Insurance Recommendation Challenge
 scale(.21174)'%3e%3cpath d='m52.487 177.086 13.018 11.852 11.762-11.762 11.896 12.266 12.846-11.358 11.827 12.43 13.339-11.013 10.897 12.728 13.27-10.145 13.947 12.319'/%3e%3ccircle cx='89.425' cy='154.584' r='5.167'/%3e%3ccircle cx='111.237' cy='156.047' r='5.167'/%3e%3ccircle cx='30.05' cy='16.5' r='4'/%3e%3c/g%3e%3c/svg%3e)
Helping Zimbabwe
$5 000 USD
Completed (~6 years ago)
Prediction
Collaborative Filtering
1791 joined
612 active
Start
01 Jul 20
Close
13 Sep 20
Reveal
13 Sep 20
Is it normal to use test data in train
Help · 7 Sep 2020, 09:13 · edited 4 minutes later · 7
Hi,
1) it normal to use test data with ones in train process, assuming that rows are already known, and will add some info in training process.
2) how handle values test values that miss from train and vice versa.
I have used Rare encoding then Wait of evidence encoding, gave me 0.08. the Rare encoding with mean aberage incoding gives the same.
If you use test data for training aswell your models are going to overfit and not generalize well.
They are various ways of dealing with missing values such as imputing by the mean or mode.
i think there is no missing values just one for date
Some occupation codes from test are missing from train and vice versia
I used 3 validation , target mean encoding and some feature enginering and get good results, with a lot of missing values in test after target encoding.
You mean from the test data. I hardly observed this
but in that case how you consider other policy values in test set out of 21 you know only 1 correct what about the rest