The World Health Organization (WHO) has shown that a lack of or uneven intake of food contributes to roughly 9% of heart attack fatalities, 11% of ischemic heart disease deaths, and 14% of gastrointestinal cancer deaths globally. More than a billion individuals are anemic due to iron deficiency (anaemia), 0.25 billion children have vitamin deficiencies ranging from vitamin A to vitamin K inadequacy, and 0.7 billion are iodine deficient, making a total of roughly 0.25 billion people anaemic. Diet recommendations are the primary goal of this study. The recommender system has to cope with a significant amount of data from the dataset in order to find relevant recommendations. In this project own data set is prepared based on various high and low values of vitamins from (vitamin a , b,c,d,e,k ) and features are divided from normal and abnormal conditions of vitamins and labels are divided in to o and 1 as normal and abnormal. Another dataset is prepared based on combination of various vitamins and their deficiency and food to be recommended based on which vitamin is deficient. In this project multiple classifier algorithms are used ( knn, decision tree, random forest, logistic regression, voting classifier ) ensembled algorithm is used to combine multiple algorithms and train a new algorithm. Accuracy of each algorithm is calculated and best algorithm is used for prediction purpose. Prediction is shown using flask web application which will detect deficiency of vitamin and recommend type of food to be taken on various combinations.