In the last chapter, we considered the most dire situation possible—biases that have been so deeply entrenched in reality that it’s impossible to collect data to refute them. Very often, however, there is the data required to keep biases out of the algorithm—but somehow the data scientist lets a bias slip through nevertheless. This chapter looks more closely at this cause of algorithmic bias.
When we discussed the model development process in Chapter 4, you learned that the data scientist not only needs to go through a lot of work steps but that many steps also require ...
