Search results

Overnight tourism in Amsterdam. A forecasting approach.

In dit artikel wordt een empirisch model gebaseerd op de aankomsten van toerisme in Amsterdam ontwikkeld en getest.

Forecasting cyanobacterial blooms using high frequency monitoring data

Cyanobacterial blooms can be toxic to humans swimming in affected waters. According to the European Bathing Water Directive bathing waters should be closed during cyanobacterial blooms. In the Netherlands, cyanobacteria monitoring in all official bathing water locations is usually performed every two weeks during the bathing season. In face of the large temporal and spatialvariability of cyanobacterial bloom dynamics this monitoring frequency however is too low for adequate early warnings to the public.High frequency monitoring and forecasting models can provide information on cyanobacterial blooms in between the regular monitoring dates and for a few days into the future. In the H2020 project EOMORES, we have combined observational data from a spectral camera (Ecowatch) near a Dutch bathing site with fluorescence data from an underwater drone to analyse the variability ofcyanobacterial blooms at short temporal and spatial scales. The results are used in a short term forecasting model of cyanobacterial blooms (AlgaeRadar) and a 3D scum forecasting model (EWACS). The AlgaeRadar is cross-validated with biweekly data from other bathing water sites and shows improved model performance compared to an earlier version that was built with only biweekly data.For the site with high-frequency chlorophyll observations the near-real time data are assimilated in the model to further enhance the model performance. Model performance of EWACS is verified using high frequency pictures from the Ecowatch station, showing scum layers on the water. This allowed us to validate and calibrate the EWACS model. Model validation abilities were in the pastalso limited by to the patchy nature and high temporal variability of the scum layers, which was not covered by sparse scum observations. With the resulting models, early warnings for cyanobacterial blooms are more reliable than those from the current practice that are merely based on biweekly monitoring data. For the protection of public health this provides better opportunities as well.

Individual intuition and communicative charisma as enablers of futures envisioning: ‘genius forecasting’ vs ‘forecasting geniuses’.

Human intent forecasting using intrinsic kinematic constraints

The performance of human-robot collaboration tasks can be improved by incorporating predictions of the human collaborator's movement intentions. These predictions allow a collaborative robot to both provide appropriate assistance and plan its own motion so it does not interfere with the human. In the specific case of human reach intent prediction, prior work has divided the task into two pieces: recognition of human activities and prediction of reach intent. In this work, we propose a joint model for simultaneous recognition of human activities and prediction of reach intent based on skeletal pose. Since future reach intent is tightly linked to the action a person is performing at present, we hypothesize that this joint model will produce better performance on the recognition and prediction tasks than past approaches. In addition, our approach incorporates a simple human kinematic model which allows us to generate features that compactly capture the reachability of objects in the environment and the motion cost to reach those objects, which we anticipate will improve performance. Experiments using the CAD-120 benchmark dataset show that both the joint modeling approach and the human kinematic features give improved F1 scores versus the previous state of the art.

Forecasting health prior to state pension age – projections for the Netherlands 2040

ObjectivesIn many Western societies, the state pension age is being raised to stimulate prolonged working. In the Netherlands, the raise of the state pension age is linked to the remaining life expectancy at age 65 with a factor of 2/3rd, and is expected to be 68 years in 2040. It is not yet well understood whether health of the 60+ permits this increase. In this study, health of Dutch adults aged 60 to 68 is forecasted up to 2040.MethodsData are from the Dutch Health Interview Survey (HIS) 1990-2017 (N≈280.000) and the Dutch Public Health Monitor (PHM) 2016 (N≈460.000). Health is operationalized using binomial scores of 1) self-rated health and 2) limitations in hearing, seeing or mobility. Categories are: good health (healthy on both items), moderate health (healthy on one item) and poor health (unhealthy on both items). First, based on the HIS, health status in 5-year age categories was modelled up to 2040 using logistic regression analysis in R. Second, the growth factor from 2016 to 2040 was applied to the health level from the PHM 2016.ResultsIn 2016, 63% of men aged 60-65 had good health, 25% had moderate health and 12% had poor health. Among women, this distribution was 64%, 22% and 14%, respectively. In 2040, the health distribution among men aged 60-68 is estimated to be 63-71% in good health, 17-28% in moderate health and 9-12% in poor health. Among women this is estimated to be 64-69%, 17-24% and 12-14%, respectively.ConclusionsHealth of Dutch cohorts nearing the state pension age in the future is estimated to remain the same or improve up to 2040. This development in health is not an obstacle to raising the state pension age. However, due to the increasing state pension age and the baby boom generation reaching age 60+ in the coming years, the absolute number of people aged 60+ in poor and moderate health that participates in labor will increase. Policy aiming at sustainable employability will therefore become increasingly important.

Time series building energy systems data imputation

Completeness of data is vital for the decision making and forecasting on Building Management Systems (BMS) as missing data can result in biased decision making down the line. This study creates a guideline for imputing the gaps in BMS datasets by comparing four methods: K Nearest Neighbour algorithm (KNN), Recurrent Neural Network (RNN), Hot Deck (HD) and Last Observation Carried Forward (LOCF). The guideline contains the best method per gap size and scales of measurement. The four selected methods are from various backgrounds and are tested on a real BMS and metereological dataset. The focus of this paper is not to impute every cell as accurately as possible but to impute trends back into the missing data. The performance is characterised by a set of criteria in order to allow the user to choose the imputation method best suited for its needs. The criteria are: Variance Error (VE) and Root Mean Squared Error (RMSE). VE has been given more weight as its ability to evaluate the imputed trend is better than RMSE. From preliminary results, it was concluded that the best K‐values for KNN are 5 for the smallest gap and 100 for the larger gaps. Using a genetic algorithm the best RNN architecture for the purpose of this paper was determined to be GatedRecurrent Units (GRU). The comparison was performed using a different training dataset than the imputation dataset. The results show no consistent link between the difference in Kurtosis or Skewness and imputation performance. The results of the experiment concluded that RNN is best for interval data and HD is best for both nominal and ratio data. There was no single method that was best for all gap sizes as it was dependent on the data to be imputed.