Customized Nutrition Recommendation with Personalized Health Profile

There are different methods which are responsible for generating the feedback for the user with the help of number and rating value. A store can sell the different kinds of items and then once the transactions of items are completed they form a stream of analysis data. The transaction data can be used to improve the sales of the store and the store do not concentrate on nutrition value to be provided by the user. In this work, an application is created with various food categories. For each category list of Food Items are shown to the end consumers from which a purchase can be made or food item description can be viewed. The purchase transaction can be made successful if the authorization cycle completes and consumer has sufficient monetary volume. In order to make suggestions to the end user, the training data is required which can be collected using a HTTP hit and then feed-backs are collected from multiple independent sources like amazons, flip-kart and mouth shut. The first step is to convert the feedback into a clean stream with removal of unwanted words, after that token queue is created, find the unique words from each feedback and then measure the count to form the frequency, from the frequency matrix a reduced matrix is created by reducing the redundancy by making use of features of the food items and removing the remaining tokens, classification of Food Item is done using single dimension k means algorithm. Each Food Item can have class label like sugar, salt, fat, energy and fat. Each of the consumer are asked series of questions and based on answers provided by the end users a nutrition score is provided which can provide a profile for the end user. The suggestions are given to the end users by making use of genetic algorithm. The user can do chatting in the application either it can be generic questions and then artificial intelligence can provide an automated answer. The chat values are Tokenization and then relationship vector is generated between the consumer and disease kind. Once the disease kind is finalized then symptoms based questions are asked in a series format and answer analysis is done to know the level of disease kind which can be Level 1, Level 2 or Level 3. If Level 1or Level 2 doctor based suggestions are provided otherwise along with suggestions links are provided related to hospital or clinic. Custom recommendations are provided based on either user kind can be BP patient, Sugar Patient or Old Age.