What is the benefit of imputing numerical or categorical features when using DT methods such as XGBoost that can handle missing values? This question is mainly for when the values are missing not at random.

An example of missing not at random categorical feature

  • Class 1: User has a red car
  • Class 2: User has a blue car
  • Class 3: User has no car (missing value)

In this case is it better to treat this feature as a binary 0/1 with NaN missing value, or as multi-label feature: 0,1 and -999 for missing?

The same question applies for a numerical feature that indicates age of user's car. Here, missing values indicate the user has no car. Is it better to keep missing values as NaN, or to impute these values? If imputing is better, should I impute using the median value and add an interaction feature for when the values are missing?

  1. it is boring and repetitive to say - so sorry - but it depends what are you trying to predict and if it is make sense to do imputation at all. If you are trying to predict the car price probably it is important to know what colour it is but if you are trying to predict whether that person will potentially default his/her loan - the car color is not relevant (unless perhaps, people who have pink cars are so rich and there is absolutely no risk to default!)

  2. is there any cost in adding new categorical feature ? e.g. why not keeping both features, both the color as well as whether the person owns a car. of course this depends on the frequency of the missing value. Later the new feature (whether the person has a car can be used to make other interactive features)

  3. regarding numerical, when it is a boosting method it does not matter because these methods inherently bin the numerical values. For example Catboost claim to have a better and more efficient binning algorithm.

  4. similar to 3 - it won't hurt if you replace the age with some really odd numbers, e.g. -99 because there is no normalization in mid-way to get skewed. The best practice is to see what percentage of the value are missing, and whether imputing will improve your performance - in short: it is very experimental.

  5. all your "meaningful" experiments will tell you retrospectively what helped. Therefore, you are highly recommended to use pipelines which help you to reproduce all steps.

  6. there are also more complicated ways of imputation - for example, making a predictive model to predict "age" in your example and then use that predicted age as an input for your second model. of course all has to be through a clean-through cross-validation to make sure there is no leakage.

  7. Go as inclusive as possible - add all sort of features generously - e.g. in your numerical value - have all medium, mean and ... and afterwards decide which made a better sense. If there was a statistician here I would have been knocked out because there is a risk of over fitting your test data ! so I hope you have an ocean of data.

  • 1
    $\begingroup$ Regarding 2): If I keep both features CAR_OWNER and CAR_COLOR, then is it better to impute the missing values in CAR_COLOR (because CAR_OWNER is FALSE) or to keep it as NaN? $\endgroup$ Jun 3 at 4:41
  • 1
    $\begingroup$ yes - you can do that - I dont think it will be called imputation - imputation is where the value is missing and you try to estimate/guess based on the neighboring values - in your example the colour_value is not missing - in fact the colour_value is invalid because there is no car ! you are not guessing the colour - you for sure, 100% know the colour is NOT_VALID - so yes, go ahead and replace it with symbolic value - the boosting algorithm will learn it easily that if the car is missing the colour will be invalid. $\endgroup$
    – user702846
    Jun 3 at 10:58
  • $\begingroup$ This makes sense. Will the boosting algorithm learn a different pattern if the missing color values are kept as NaN instead of being replaced by symbolic value? $\endgroup$ Jun 3 at 14:25
  • $\begingroup$ now that I think - I think you should replace NaN with symbols - e.g consider these cases, Car A has an exotic color (e.g. cyan) which was not within the define official colors for the operator to put - or it was colored in a different country - versus the case where there is no color. Or perhaps the owner has changed the color but through the admin work has been missed. So def. there is a difference btw missing color and "not-available" color. $\endgroup$
    – user702846
    Jun 3 at 21:25
  • $\begingroup$ Regarding whether the boosting algorithm learn a different pattern - it depends on the data and out outcome variable. but you can always visualize the decision trees e.g. towardsdatascience.com/… - this helps you to design better features an see what features are redundant. $\endgroup$
    – user702846
    Jun 3 at 21:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.