Firstly, lets suppose model omits the 'Size' as most significant feature, so what is implied here, having larger size or lower size of an app contribute to the rating? What If there is no ascending or descending order in the attribute, for instance, if the 'Category' is most significant, then what category contributed the most?
Decision Tree splits the ...
You can frame this issue as feature importance. Which features have the greatest influence on the target value of churn rate?
There are many ways to approach feature importance. In decision trees, permutation importance can be used.
As stated this makes no sense to train the same model on different samples and compare the results.
If you have only one problem to solve, and one model, you should only have one data set,
and emphasis on data preparation and feature enginering.
While featrue enginering or running the model,
you should detect which features are the moste important ones.
Random Forest can work good here since it is a decision tree.
Before you call RandomForest, you will have to OneHotEncode your categorical variables e.g. Butter Flour Eggs ... because the Regressor or Classifier (whichever you fancy) cannot work with string and NaN values.
The number of features can be used to handle two situations:
High bias (the common one): Adding features is one way to approach models with high bias because additional features can increase the predictive power of your data. This is commonly done as part of feature engineering.
High variance (the uncommon one): However, a large number of features can also ...
The method proposed seems to be related to the topic of Confident Learning/noisy labels. Check these out:
Confident Learning: Estimating Uncertainty in Dataset Labels,
In that setting you can either remove datapoints or change their label, but on the validation set.
I have the same problem, and I'm considering to do the same approach: delete the extremely bad predicted data points. I investigated the bad predictions and I came up with two results:
-Either the data are not correct, there has been an error when recording them
-Or there is an important attribute that is not presented in the dataset and is responsible for ...
First, you need to understand what Sequential Modeling is?
There are two categories in which Sequential Modeling falls in
Suppose the data itself is a Sequence of the stream like audio, time-series data, textual data.
Other is if you have the model who is working in sequential manner, what it means? Suppose you are giving a model training data but the model ...