Current novice-friendly machine studying (ML) modeling instruments focus on a solo person expertise, the place a single person collects solely their very own knowledge to construct a mannequin. Nevertheless, solo modeling experiences restrict priceless alternatives for encountering different concepts and approaches that may come up when learners work collectively; consequently, it usually precludes encountering important points in ML round knowledge illustration and variety that may floor when completely different views are manifested in a group-constructed knowledge set. To handle this concern, we created Co-ML – a tablet-based app for learners to collaboratively construct ML picture classifiers via an end-to-end, iterative model-building course of. On this paper, we illustrate the feasibility and potential richness of collaborative modeling by presenting an in-depth case research of a household (two youngsters 11 and 14-years-old working with their dad and mom) utilizing Co-ML in a facilitated introductory ML exercise at dwelling. We share the Co-ML system design and contribute a dialogue of how utilizing Co-ML in a collaborative exercise enabled rookies to collectively interact with dataset design concerns underrepresented in prior work akin to knowledge variety, class imbalance, and knowledge high quality. We focus on how a distributed collaborative course of, through which people can tackle completely different model-building tasks, offers a wealthy context for kids and adults to study ML dataset design.