This essay explores how to broaden the scope of what constitutes anthropological and ethnographic research by cross‐fertilizing with data science. I discuss four types of relationships anthropologists have sought to foster with data science: anthropology of data science, anthropology over data science, anthropology with data science, and, the least developed of the four, anthropology by data science. Data science as a field has cultivated abductive, bottom‐up forms of quantitative research, which provide useful quantitative parallels to similarly abductive, bottom‐up qualitative techniques in ethnographic research. Anthropologists should adopt an anthropology by data science perspective through incorporating machine‐learning and other data science techniques into anthropological research. Nick Seaver's concept of bastard disciplines and methodologies provides a helpful framework for such work. [data science, machine learning, ethnography, anthropology]