Learning from data with structured missingness

Missing data are an unavoidable complication in many machine learning tasks. When ‘missing at random’ there exist a range of tools and techniques to deal with the issue. However, as studies become more ambitious, seek learn from ever-larger volumes heterogeneous data, increasingly encountered problem arises which missing values exhibit association or structure, either explicitly implicitly. Such ‘structured missingness’ raises challenges that have not yet been systematically addressed, presents fundamental hindrance scale. Here we outline current literature propose set grand structured missingness. Gathering big datasets has essential component scientific areas, but it is some missing. An important growing effect needs careful attention, especially when sources combined, missingness, where random, specific structure.