The agriculture-food systems in low-and middle-income countries like India change rapidly and so cause the co-existence of three forms of child malnutrition: under-nutrition, micronutrient deficiencies, and diet-related overweight. This triple burden of malnutrition complicates aid-decision-makers’ choice of suitable policies and investments in agriculture and food systems for improved nutrition and health that would benefit most children in a district. The goal of this proposal is to guide aid decision-making in a twofold way: by selecting the most crucial determinants of the Indian agriculture-food system that drives child malnutrition at the district and by generating an updatable prediction tool for the three different types of child malnutrition. I use household survey and geospatial data from AReNA’s DHS-GIS Database, the Demographic and Health Survey in India (2019/20), and other geospatial, freely accessible, and frequently updated data bases. In the analysis, I will select the determinates in the Indian agriculture-food system with the greatest predictive power for the prevalence of the different types of child malnutrition at the district level and I will model the prediction tool. I will employ two different approaches - either machine learning or a common regression technique. In a final step I will compare the prediction accuracy of the two models. The innovations of this proposal are the use of machine learning techniques for mapping child malnutrition at the district level in India and the creation of an updatable tool for different types of malnutrition with the possibility to interpret the importance of single determinants in agriculture-food systems.