Atopic diseases (asthma, eczema, rhinitis) are heterogeneous, both in their course (curricular heterogeneity, i.e., differences in the time-course of their development) and their cause (aetiological heterogeneity, i.e., differences in the underpinning pathological mechanisms). Despite this diversity of mechanisms and later outcomes, at symptom onset, clinical presentation in different subgroups of patients (such as transient or persistent wheezers) may be similar. Consequently, it is difficult to predict the course of allergies in individual patients, and the mechanisms associated with the persistence and remission of symptoms are poorly understood. Collection of vast amounts of longitudinal clinical data and high-throughput technologies may help us to understand the complexity of the development of childhood allergies. However, it is challenging to process, analyze, and interpret the large volumes of clinical and biological data. Machine Learning (ML) is a fundamental technology which may meaningfully process data that exceed the capacity of the human brain to comprehend, and ML models can digest large amounts of data quickly and identify underlying patterns within large data sets. However, it is important to emphasize that these patterns do not necessarily correspond to underlying biologic pathways. This chapter provides examples on how different ML approaches and methodologies (predictive and explanatory) have been used in pediatric allergy research, focussing on two specific exemplars: (1) understanding heterogeneity of childhood wheezing illness, including prediction models for childhood asthma and (2) understanding heterogeneity of childhood allergic sensitization and its relationship with asthma.