Data-driven customer acceptance for attended home delivery.

Home delivery services require the attendance of the customer during delivery. Hence, retailers and customers mutually agree on a delivery time window in the booking process. However, when a customer requests a time window, it is not clear how much accepting the ongoing request significantly reduces the availability of time windows for future customers. In this paper, we explore using historical order data to manage scarce delivery capacities efficiently. We propose a sampling-based customer acceptance approach that is fed with different combinations of these data to assess the impact of the current request on route efficiency and the ability to accept future requests. We propose a data-science process to investigate the best use of historical order data in terms of recency and amount of sampling data. We identify features that help to improve the acceptance decision as well as the retailer's revenue. We demonstrate our approach with large amounts of real historical order data from two cities served by an online grocery in Germany.

Assessment of transportation system disruption and accessibility to critical amenities during flooding: Iowa case study.

Natural disasters, such as flooding, can cause severe social, environmental, and economic damage to a community. Transportation infrastructure plays an essential role in flood response and recovery efforts. However, flooding may disturb road functionality and generate direct and indirect adverse impacts, including the loss of access to essential services. This paper presents a comprehensive analysis of flood impacts on road network topology and accessibility to amenities for major communities in the State of Iowa using graph-theoretic methods, including single-source shortest path analyses. We assessed the disruption of transportation networks on the accessibility to critical amenities (e.g., hospitals) under 100 and 500-year flood scenarios. Our analysis methodology leads toward the development of an integrated real-time decision support system that will allow decision-makers to explore "what if" flood scenarios to identify vulnerable areas and population in their authority. These analyses could promote possible improvements (e.g., temporary relocation of critical services) to mitigate the consequences of road system failure during flooding. Due to varying environmental conditions at specific locations and effects on road topology under flood events, the results show differential impacts in edge and node losses as well as access to critical services. Results indicate that floods can lead to edge losses of up to 18%, and not only large cities but also some small cities can experience significant vulnerability to flooding. Some new or reconstructed bridges have failed to operate during analyzed flood events. During the 100 and 500-year flood return periods, the total number of inaccessible bridges within the selected cities is 184 and 294, respectively. Our work found that the shortest path length to the closest critical amenity under baseline condition can flip to the second or higher-orders during flooding. Many critical amenities have been found at risk of flooding in the studied cities.

Prediction of ultrasound-mediated disruption of cell membranes using machine learning techniques and statistical analysis of acoustic spectra.

Although biological effects of ultrasound must be avoided for safe diagnostic applications, ultrasound's ability to disrupt cell membranes has attracted interest as a method to facilitate drug and gene delivery. This paper seeks to develop "prediction rules" for predicting the degree of cell membrane disruption based on specified ultrasound parameters and measured acoustic signals. Three techniques for generating prediction rules (regression analysis, classification trees and discriminant analysis) are applied to data obtained from a sequence of experiments on bovine red blood cells. For each experiment, the data consist of four ultrasound parameters, acoustic measurements at 400 frequencies, and a measure of cell membrane disruption. To avoid over-training, various combinations of the 404 predictor variables are used when applying the rule generation methods. The results indicate that the variable combination consisting of ultrasound exposure time and acoustic signals measured at the driving frequency and its higher harmonics yields the best rule for all three rule generation methods. The methods used for deriving the prediction rules are broadly applicable, and could be used to develop prediciton rules in other scenarios involving different cell types or tissues. These rules and the methods used to derive them could be used for real-time feedback about ultrasound's biological effects.