Workspace Mobile Application Finder Based on Machine Learning Approach

Abstract

Finding the best-fit workspace – collection of heterogeneous places to meet, work or study – is considered a problem for clients who cannot determine the availability of the spaces at a specific moment. The availability of spaces in workspaces is another problem because it needs real time communication with each workspace. Therefore, a system that searches in a registered workspace to find the nearest available workspace using a mobile app developed by flutter. A backend Laravel system that communicates with the workspace software to find a real time state is introduced in this paper. In addition, software is built that manages workspace and keeps track of its availability. The main merit of this software is accepting the request of the first solution software to give a real time state of the available spaces in its workspace. Discern a workspace to fall under a certain label allows us to recommend workspaces to clients easily based on machine learning ML) techniques.Our main approach involved is using traditional ML methods, where we train and build a model based on a dataset and use that model in order to predict a rating that a client might give a workcspace. A client is able to give a rating from 1 to 5 stars meaning we have 5 separate labels that we need to classify a workspace under. Since we need to predict a discrete label from 1 to 5, this is a classification problem. So, therefore a classification-based approach, where we use ML models' classification in order to classify a workspace between 5 labels each representing a possible rating that a client could rate a workspace. Furthermore, separate applications that can communicate are developed in this paper. In addition, a platform is performed that offers both workspaces and clients a means of interacting with each other. In this paper, a system that records all workspace-related data, including name, location, and lists of available workers, is constructed. This gives customers a way to register and reserve a space or a seat at the selected workspace as well as a way to keep track of all their interactions. It is observed that by generating a synthetic dataset that represent the relation between certain features and rating and by using a random forest model, we are able to predict the rating group of a workspace. This is achieved utilizing ML techniques with a classification accuracy of 0.79 and a precision and recall values of 0.787 and 0.790, respectively, giving much better scores compared to other models we tested.