Summary

In this chapter, you learned the core data science skill of imputation. Imputation allows the replacement of missing values for features with educated guesses as to what the real values are. Imputation is necessary to provide predictions for all samples, including those with missing feature values. This is because many machine learning algorithms, including those we've worked with in this book, cannot take input which includes missing values. Imputation methods range from simple ones, such as imputing with the mean, median, or mode of non-missing values, to more complex methods, including creating a predictive model for the feature that has missing values.

After ensuring that the model we deliver can provide predictions for all samples, via imputation, we conducted a financial analysis. While we left this to the end of the book, an understanding of the costs and savings going along with the decisions to be guided by the model should be understood from the beginning of a typical project...