Discussion of Results. Representation and coverage
Discussion of Results. From the presented findings, the following high-level conclusions can be drawn from our user-focussed problem discovery observations. When combined, both UMUX and UEQ post-task data reveal important data on how effective the simulation tool was and the overall users’ experiences when completing crowd simulation tasks in Unity. The overall usability of the crowd simulation software was considered acceptable, with an average UMUX score of M = 68.06 (SD = 11.20) supporting this claim. Furthermore, the initial analyses of the UEQ indicated that the tool was attractive to use (M = 1.00; SD = 1.06), with positive evaluations for both hedonic (M = 0.63) and pragmatic (M = 1.25) qualities. Unpacking the lower hedonic qualities may further reveal the underlying reasons surrounding these evaluations and a discussion of these results is presented. To begin with, the pragmatic aspects, or task-related quality of the tool, are detailed. The “efficiency” score (M = 0.79; SD = 0.87) reflected well on the cognitive resources demanded by the cohort when concerning the accuracy and completeness of the goals they had achieved during the different tasks. Furthermore, the users generally rated “perspicuity” highly (M = 0.83; SD = 0.93), indicating that they felt that the crowd simulation software was somewhat easy for them to become familiar with. This also suggested that it was easy for the users to learn how to use the software in the short amount of time they were given as well as the simulation methodology that was undertaken. In comparison, the UEQ measure of “dependability” was rated relatively low (M = 0.25; SD = 1.04), undermining the idea that the users felt that they were in control and suggesting that they felt insecure using the tool when undertaking the simulation tasks, as well as indicating that the system was perceptually unstable or behaved in an unpredictable way. This was true for all participants as the Alpha-Coefficient (ɑ = 0.82) suggested consistent measures. The hedonic, or non-task related qualities, user ratings of “stimulation” (M = 1.71; SD = 0.80) indicated that the extent to which the tool provided users with innovative and interesting functions, interactions, and stimulation was well received. The evaluation of “Novelty” (M = 0.79; SD = 1.11) also served to represent how innovative the cohort considered the crowd simulation tool to be. Therefore, due to the “newness” of the simulation tool, the novelty of the software was to be a compounding fact...
Discussion of Results. Significant results obtained from the analysis of the coldchain update survey are as follows (refer to Table 1 on following page): From the Cold Chain Update Survey, the number of villages served have increased by 35 % from the basic cold chain survey Also the total number of villages lacking an Integrated Health Service Posts (Posyandu) has decreased. The proportion decreased from 19.5% to 3.7%. Changes in staff were minimal. Although there decrease in the number of mid-wives and immunizers, this did not seem to affect the performance of duties and delivery of services.
Discussion of Results. At the time of the consultation following the evaluation, Xx. Xxxxxxx will provide you with a very thorough written report detailing all results, a diagnostic impression and summary, a formal diagnosis, and several recommendations. Xx. Xxxxxxx will discuss with you (patient or parent of the patient) her working understanding of the problem, treatment plan, and her view of the possible outcomes of treatment. For adolescent patients it is frequently preferable for Xx. Xxxxxxx to have separate consultation with parents and the adolescent. Adolescents frequently are more open to hearing results and recommendations when they are in private consultation with Xx. Xxxxxxx. Additionally, separate consultations allow for more candid discussions between Xx. Xxxxxxx and parents. Patients who are 18 years of age must provide consent for parents to be informed of the results of the evaluation.
Discussion of Results. The communication overhead experiments indicate a significant difference be- tween active and passive mode with regard number of messages per minute and the total network traffic generated, as can be seen in Figure 5. Passive mode clearly uses fewer messages per minute than active mode. This xxx- xxxxxxx can be clarified with an explanation of the inner workings of passive mode. When no conflict mediation is requested, communication occurs di- rectly between consumer and provider. No TTP or TMM are involved. This results in fewer messages per minute than the corresponding active mode. The difference in the network traffic generated by active and passive mode can be seen in Figure 6. Passive mode clearly generates significantly more bytes per minute than active mode, especially when conflict mediation is re- quested. In the case that no conflict mediation is requested, consumer and provider exchange cryptographically encrypted portions of their individual audit logs. These audit logs are much larger than the simple metrics ex- changed in active mode. In the case that conflict resolution is requested, all parties send their entire audit log to the TTP, resulting in a substantial increase in network traffic.
Discussion of Results. The scaling results shown in Figure 7 indicate that both monitoring modes can scale, in terms of communication overhead. Doubling the number of agents doubles the number of messages. The distributed scaling experiments 64 256 448 640 832 1024 1216 1408 1600 1792 1984 2176 2368 2560 2752 2944 3136 3328 3520 3712 3904 4096 Figure 8: Average CPU load of large scale monitoring on DAS-4. shown in Figure 8 indicate that both monitoring modes can scale in large, distributed systems with only minimal computational overhead. In active mode, CPU load is affected by the interval of measurement. In passive mode, CPU load is determined by both the interval of measurement and the frequency with which mediation is requested. The passive mode has a higher average CPU load than the active mode. This is to be expected, given that passive mode requires CPU-intensive cryp- tographic computations4. Active mode uses no cryptography, because secu- rity is guaranteed by the TTP. What cannot be seen in these particular figures is the level of interaction with the TTP. In passive mode, there is only a minimal amount of interac- tion with the TTP (e.g. exchanging cryptographic keys). All computations, including measurements and cryptography, are performed by the consumer and provider. In contrast, active mode relies on the TTP to perform all measurements. With this in mind, the active mode shown in Figure 8 essen- tially depicts the CPU load at the TTP. The corresponding CPU load at the consumer and provider is insignificant. In contrast, the passive mode shown in the same figure essentially depicts the average CPU load at the consumer and provider. The corresponding CPU load at the TTP is insignificant. In a 4In particular, the BLS implementation is chosen only for its functionality. This code is not optimized for production level systems. ‘real world’ environment, a party would likely pay for the services of a TTP. In this case, the reduction in TTP interaction associated with passive mode would result in reduced financial costs.
Discussion of Results. This experiment demonstrates how monitoring overhead reflects perceived risk. The self-adaptive monitoring policy used for this experiments starts with a high level of perceived risk. After a period of time without incident, the level of perceived risk lowers. This is reflected by increasing the measurement interval and thus reducing overhead. Eventually, this leads to a switch to passive mode, which reduces dependence on a TTP. In the case that a violation is detected, mediation is requested. This may result in a higher level of perceived risk and a correspondingly high level of monitoring. This significantly increases the message and CPU overhead but offers higher assurance. By adjusting the monitor to match the level of perceived risk, overhead can be substantially reduced when possible.
Discussion of Results. Results displayed in the table in Section D show that Plan’s KIDCARE Child Survival Project met, and significantly exceeded population based coverage targets in most project objectives, including several proven high-impact child survival activities. In addition, health system capacity strengthening and emphasis on sustainable systems have left the Kilifi DMOH in the position to serve as a continuing “Learning Center” for other DMOH in Kenya. KIDCARE also subjected the final evaluation results to the lives saved calculator. The project saved at least 989 lives with up to 396 in its final year. There is an estimated drop of Under 5 Mortality Rate of 31% from baseline levels. The greatest success at saving lives was achieved for Malaria and Measles. In spite of significant improvements in several MCH indicators neonatal causes, HIV/AIDS, pneumonia and diarrhea continue to be major causes of childhood deaths in the area. The following graphs compare selected high-impact child survival indicators with preliminary findings from the 2008-2009 DHS. The full report will not be available until early 2010. Province level data is not yet available for all indicators, so national rural results are provided in some cases. Coast Province includes one major (Mombasa) and several smaller urban and periurban areas, whereas the KIDCARE project was implemented in the rural areas only. Data should be interpreted with caution since sample sizes and methodology in surveys are not exactly comparable. Even the DHS report states that samples in any given Province are small. Nevertheless, it is helpful to compare findings within the same population over time. There were other major maternal/child, HIV/AIDS and malaria activities going on in Kenya and in the District at the same time. The Kilifi DMOH attributed the extent and magnitude of improvements in the indicators that were recognized at the national level, were the direct result of the assistance provided by KIDCARE project. 60 50 40 Series1 30 20 10 0 K D CARE 2005 K D CARE MTE K D CARE Final 2009 DHS 2008 National* 60 50 40 Series1 30 20 10 K D CARE 2005 KID CARE MTE KID CARE Final 2009 DHS 2008 National* 60 50 40 Series1 30 20 10 0 K D CARE 2005 K D CARE MTE K D CARE Final 2009 DHS 2008 National* Contribution toward Objectives The KIDCARE Project used variations of two complementary strategies to reach project objectives: development of a community based health system with strong links to MoH service providers, and designing...
Discussion of Results. The non-linear processes behind wind and non-wind damage, as well as the effects of cascading failure of structural components, entail that reduced-form approaches as discussed here may only approximate the actual storm damage characteristics. In order to assess the robustness and quality of macroscale storm damage functions, we have analysed and compared the results of four different models applicable to the European winter storm season. for both the calibration and the evaluation of damage functions. With a focus on the level of extreme losses, least-squares curve fitting has often been employed to calibrate damage curves to loss data. The combination of skewed loss distribution and het- eroscedastic variance seen for the case of GDV data suggests a violation of the basic assumption for least-squares fitting and potentially leads to biased results. Due to the high dynamic range even temporally or spatially aggregated loss figures, as used in the cubic excess-over-threshold damage function by Klawa & Xxxxxxx (2003) [model K], are subject to this effect as they are still dominated by extreme losses. The optimal curve fitting procedure remains a matter of discussion. Relying on the assumption of general damage relation valid for a large range of losses, the probabilistic power law damage function by Xxxxx et al. (2012) [model P] puts equal weight on all data points. In contrast, the fitting procedure for the probabilistic claim-based damage function by Xxxxxx & Ruck (2008) [model H] has given greater weight to extremes by using averages of binned losses. The comparison between model H and the simple exponential damage function [model X], both of which are calibrated in the same manner, shows that effective calibration relies on a combination of model constraints and curve fitting. As was seen in Fig. 23, model H attains greater robustness against jackknife variations of the training sample due to the presumption of a specific claims distribution. Following a different philosophy, model P achieves robustness by rooting the damage function in the entire range of loss. Transferability is one of the biggest challenges of empirical damage functions. All of the discussed damage functions require substantial calibration to loss data. On the one hand, Xxxxxx & Xxxxxxx (2011) applied their damage function to Germany by assuming a parametrisation unchanged to that of the federal state of Baden-Württemberg. Donat, Leckebusch, Wild & Xxxxxxx (2011), on the other hand, ...
