Optimization of Single Mode Trip: Technique-Based Recommendation System of Machine Learning

EasyChair Preprint 6753

Abstract

Smart city is a recent topic, but it is developing very rapidly, as it is seen as a winning strategy to deal
with some serious urban problems such as traffic, pollution, energy consumption, waste treatment.
Mobility is one of the most difficult subjects to address. It contains both environmental and economic
aspects and requires both high technology and virtuous human behaviour. Intelligent mobility is largely
permeated by ICT, used in upstream and downstream applications, to support the optimisation of traffic
flows, but also to gather citizens' opinions on the quality of life in cities or the quality of transport
services.
In this context, the present brief aims to develop a system for recommending the best routes for
passengers according to their departure and arrival addresses.
To meet this objective, it is necessary to carry out an analysis in order to define the different tools and
methods to be used. In addition, after the identification of user behaviour needs. We carried out a
design
adequate to our recommendation system and well detailed for each module. Then we compared the
different models RandomForest, artificial neural networks, KNN Basic, KNN Means, KNN ZScore,
SVD. Finally, we found that the two models RandomForest and artificial neural networks are the most
efficient compared to the other models, with an accuracy of 0.97 for the first one and 0.90 for the
second one.

Keyphrases: Artificial Neural Networks, Recommendation System, deep learning, intelligent mobility, machine learning

@booklet{EasyChair:6753,
  author    = {Chaimae Attar and Najima Daoudi},
  title     = {Optimization of Single Mode Trip: Technique-Based Recommendation System of Machine Learning},
  howpublished = {EasyChair Preprint 6753},
  year      = {EasyChair, 2021}}