The t-test, resource pooling, and psi: why the minimum sample size counts towards theoretical progress in behavioral science

A theoretical construct that subsumes an empirical phenomenon should rest on statistically significant test-results with high replication probably. To statistically establish such test-results, behavioral science publications typically rely on a t-test, and researchers typically operate under limited data collection resources. To publish more test-results, or to publish an individual result sooner, questionable strategies are commonly used to reduce data collection cost. One strategy is to increase the β-error rate from 0.05 to 0.20; a second strategy is to treat the control group as a constant, collapsing a two sample t-test into a one sample t-test. Both strategies happen to underlie Bem’s (2011) rightly controversial results on human precognition (“psi hypothesis” [Journal of Personality and Social Psychology, 100, 3, 407-425]). Since both strategies undermine theoretical research in behavioral science, their ubiquity partially explains the rarity of well-corroborated theoretical constructs there. We advocate collaboration between individual labs as a viable route to enabling theoretical research by collecting a large enough sample jointly.

3D car shape reconstruction from a contour sketch using GAN and lazy learning

Abstract3D car models are heavily used in computer games, visual effects, and even automotive designs. As a result, producing such models with minimal labour costs is increasingly more important. To tackle the challenge, we propose a novel system to reconstruct a 3D car using a single sketch image. The system learns from a synthetic database of 3D car models and their corresponding 2D contour sketches and segmentation masks, allowing effective training with minimal data collection cost. The core of the system is a machine learning pipeline that combines the use of a generative adversarial network (GAN) and lazy learning. GAN, being a deep learning method, is capable of modelling complicated data distributions, enabling the effective modelling of a large variety of cars. Its major weakness is that as a global method, modelling the fine details in the local region is challenging. Lazy learning works well to preserve local features by generating a local subspace with relevant data samples. We demonstrate that the combined use of GAN and lazy learning produces is able to produce high-quality results, in which different types of cars with complicated local features can be generated effectively with a single sketch. Our method outperforms existing ones using other machine learning structures such as the variational autoencoder.

A SPATIAL-AND-SCALE-DEPENDANT MODEL FOR PREDICTING MSW GENERATION, DIVERSION AND COLLECTION COST BASED ON DWELLING-TYPE DISTRIBUTION

Comprehensive models were developed to predict waste generation for different collection streams. Taking into account the dwelling-type distribution encountered during the different waste collections, it was possible to better capture the waste generation variability. Using the same approach, collection and transportation cost models were also developed. This series of models were validated using data from the Urban Agglomeration of Montreal (UAM), which is composed of 33 districts with widely different scales of population and dwelling characteristics. The unknown parameters of the models were identified through mean square regressions applied on the real data available for the case-study. For example, values of 1.364, 1.019 and 0.500 t/(dwelling.yr) were identified for the total quantity of wastes generated in single-family, duplex and other dwelling, respectively. Using the same approach, it was possible to determine collection time as a function of the dwelling-type distribution along the collection route. Values of 28.7 s, 11.4 s and 5.22 s were identified as the collection time per dwelling for single-family, duplex and other dwelling, respectively. Equipped with a combination of fitted parameters and reported values from the literature, the models were used as predictive tools. Three features are illustrated in this paper: 1) the simulation of various scales for the generation, diversion and specific collection cost; 2) the effect of adding a new collection stream; 3) the effect of an increase of the citizen participation to a specific collection stream. Predicted results enable decision-makers to have access to very useful information.

On the influence of collection cost on reverse channel configuration

This paper revisits the impact of collection cost on a manufacturer’s optimal reverse channel choice. A manufacturer who remanufactures his own products has the choice between managing collection of used products himself, let the retailer manage collection or involve a third party company to manage collection. In particular, we consider a convex collection cost function depending on the collection rate. Contrary to previous literature, we show that the manufacturer always prefers retailer-managed collection, independent of collection cost. The retailer will always choose a positive collection rate. If collection cost is above a certain threshold, not all used products will be collected and the manufacturer (almost) collects all channel profits. Third party-managed collection is always dominated. In extensions, we also consider a restriction to equilibria and a minimum collection rate, which may be imposed by regulation. Both extensions may change the reverse channel choice to manufacturer-managed. Moreover, we see that it may be impossible for regulation to increase collection because the profit-maximizing collection rate may already be the highest economically viable one.

Innovative Solutions for Local Public Administration in the Process of the Waste Collection

Sustainable development has always been one of the most important policies implemented by the European Union, mentioned in different treaties over time. In 2015, European Union is setting 17 Sustainable Development Goals (SDG) with 169 targets within the 2030 Agenda for Sustainable Developement. European Union is preseting the SDG 11 Goal aim ”to renew and plan cities and other human settlements in a way that they offer opportunities for all, with access to servicies, energy, housing, transportation, green public spaces, while improving resource use and reducing environmental impacts”. The EU is monitoring the progress towards SDGs for all member states through different indicators, like the recycling rate of municipal waste for SDG 11. In 2018, Romania has the recyling rate of municipal waste of only 11,11%, far away from the average of EU of 47,4%. The local public administrations have an important role in increasing the recyling rate, by providing the best services for the citizens and, in the same time, by reducing the costs of these services. In this matter, the paper presents an innovative solution regarding the waste collection in the largest District from Bucharest, Romania. The underground waste collection platforms represent and alternative for the classic platforms and the innovation comes from the smart component that they integrate, represented by the filling sensors that communicate directly to the sanitation operator. The monitoring system has direct effect by incresing the efficiency of the waste collections process, as follows: reduces the waste collection cost by 50%; improves the services provided for the citizens; predicts waste generation patterns based on data; refines waste collection processes over time; optimizes routing and navigation etc.

Investigation of Home Delivery Models and Logistics Services in China

The last-mile issue is of great concern in coping with the considerable development of the internet shopping market in China. Several issues arise from home delivery activities for fulfilling internet shopping orders; for example, increased operating costs for handling failed home deliveries, and deteriorating traffic conditions resulting from frequent delivery trips. To improve the logistics efficiency of home delivery operations and solve the problem of delivery failures, pick-up points (PPs) and self-delivery boxes (SDBs) are being implemented in China. This study investigates three home delivery models including the traditional model, the PP model, and the SDB model. Under each model, the carrier’s delivery distance and the customer’s collection distance are calculated. According to the distance, the costs of the three delivery models are compared. To simulate the carrier’s delivery route, an ant colony algorithm combined with genetic algorithm is developed to optimize the delivery route in this research. The research findings are: (1) Both the PP model and the SDB model are capable of reducing the customer’s collection cost significantly, by between 29.1% and 84%, when at least 30% of home deliveries are missed. (2) The SDB model is more favorable in relation to reducing the delivery costs of the express company, by between 67.1% and 71.3% when the proportion of missed home deliveries ranges from 20% to 50%. (3) Among PPs using the post office, convenience store, and subway station, the subway station network is the most effective scheme in relation to reducing the customer’s collection cost, by 84%.

Optimizing Field Data Collection for Individual Tree Attribute Predictions Using Active Learning Methods

Light detection and ranging (lidar) data are nowadays a standard data source in studies related to forest ecology and environmental mapping. Medium/high point density lidar data allow to automatically detect individual tree crowns (ITCs), and they provide useful information to predict stem diameter and aboveground biomass of each tree represented by a detected ITC. However, acquisition of field data is necessary for the construction of prediction models that relate field data to lidar data and for validation of such models. When working at ITC level, field data collection is often expensive and time-consuming as accurate tree positions are needed. Active learning (AL) can be very useful in this context as it helps to select the optimal field trees to be measured, reducing the field data collection cost. In this study, we propose a new method of AL for regression based on the minimization of the field data collection cost in terms of distance to navigate between field sample trees, and accuracy in terms of root mean square error of the predictions. The developed method is applied to the prediction of diameter at breast heights (DBH) and aboveground biomass (AGB) of individual trees by using their height and crown diameter as independent variables and support vector regression. The proposed method was tested on two boreal forest datasets, and the obtained results show the effectiveness of the proposed selecting strategy to provide substantial improvements over the different iterations compared to a random selection. The obtained RMSE of DBH/AGB for the first dataset was 5.09 cm/95.5 kg with a cost equal to 8256/6173 m by using the proposed multi-objective method of selection. However, by using a random selection, the RMSE was 5.20 cm/102.1 kg with a cost equal to 28,391/30,086 m. The proposed approach can be efficient in order to get more accurate predictions with smaller costs, especially when a large forest area with no previous field data is subject to inventory and analysis.