Explainable AI using Data Robot to Identify the Nutrition Values of Ordered Food

English

Abstract

Explainable artificial intelligence (XAI) is an advanced form of machine learning that allows people to understand and trust machine learning algorithms' results. Users can drag and drop the dataset to test their model, which can be completed in about an hour. This paper aims to investigate and simulate the explainable AI using a Data robot to identify the nutrition value of ordered food by predicting the health condition features. The framework was created automatically in the data robot system and could select the target of your searching dataset to give the best model. This research will use a data robot as a platform. The reason the data robot was chosen is that it can use for making automation predictions and fitting graphs that show a certain detail in terms of an outlier, and visualizations that can easily interpret the result more understanding. Besides, data robots could assist you with a bunch of machine learning models that can be used for test prediction and deployment. The model is organized well in the data robot after the dataset has been implemented along with the target prediction. The performance of the metric is also recommended by the data robot that fits with the dataset itself. In this experiment, two types of models that work on binary classification have been selected named as Keras Slim Residual Neural Network Classifier using the Training Schedule Model and Generalized Additive2 Model. It would be comparing the performance accuracy at the end of the deployment of the dataset. It goes to show which model used has a better performance to explain the nutrition value of ordered food by predicting the health condition and could be implemented in the food industry in the future.