1: \begin{abstract}

2: %    We develop a fully non-parametric, fast, easy-to-use and powerful test for %missing completely at random (MCAR) missingness, using Random Forest. The test compares distributions of different missing pattern groups on random projections of the data and is, in principle, able to detect arbitrary distributional differences on these projections. The use of random projections makes it applicable, even if very little or no fully observed observations are available and allows to extract information complementing the rejection decision. An efficient permutation approach guarantees the level for any finite sample size, resolving a major shortcoming of most other available tests. Finally, the test can be used on both discrete and continuous data and scales well in the number of dimensions. %\textcolor{red}{Add some more sentences how amazing the test is}

3: %\end{abstract}

4: