Vehicle routing problems (VRP) and Container Loading Problems (CLP) have been studied for decades. However, the combination of the two deserves more attention than in the literature to date. When solving VRP problems, computed routes must be checked for feasibility. Among the feasibility checks to perform, we need to guarantee that the load plan is feasible, namely that all the assigned products fit inside the truck. This involves solving a CLP.

Since the check of load plan feasibility is performed frequently, a short computational time is important. Hence, the load plan feasibility check is usually performed using approximation methods. Having rapid and reliable load plan feasibility estimations is crucial to reduce computational times when solving the VRP problem. However, if these estimations are conservative, the obtained routes are inefficient routes; if the estimations are opportunistic, the resulting load plans can turn out to be infeasible.

This work explored to what extent supervised Machine Learning (ML) methods can be used to rapidly yet accurately classify whether load plans will be feasible or not. These predictions can then be exploited in VRP algorithms to improve efficiency and computation time.

Several ML methods are considered and benchmarked on synthetic data and real
data from a major company in the beverage sector. Extended experiments in different settings are performed, to check the effectiveness of ML in providing reliable load plan estimations and to extract insights on how load plan characteristics affect load feasibility.

Results suggest the effectiveness of applying ML models, with Random Forest models reaching an accuracy above 91.5% on all different experiments considered. Also, compared to the current estimations used for load feasibility checking, Random Forest models decreases computation time with 54.9%.
... Read more

The department of Urbanism at the TU Delft, our clients, research the sentiment in different places, times, ages or genders and compare them to each other. This report describes the purpose, design, implementation and accuracy of a web tool created to get insights into the sentiment people have, de- ducted from social media. The aim of our project was to make research easy and extracting innovative insights from social media.

In our tool, we analysed tweets and their location to collect information about the sentiment different people have towards places. The implementation of our tool consisted of five main components: (1) Twitter-Kafka, processing the tweets from the tweet data stream to our database, (2) face recognition, used for determining whether a tweet comes from a person instead of a company or organisation and for age and gender inference, (3) sentiment analysis, using machine learning to determine whether a tweet is neutral, negative or positive, (4) REST API, for the connection between the front-end and the back-end and (5) the user interface, in the form of an interactive dashboard.

At the beginning of the project, we set up a pipeline that checks the code on multiple things. The testing of the back-end is based on a Python unit test suit. For the build to succeed, all tests must pass and the total branch coverage must be at least 80%. We used Flake8 and ESlint in our build to ensure code quality at all times.

All of the above-mentioned components are up and running. The clients are now able to research the sentiments of people towards places. ... Read more