Smoothing Spline Gaussian Regression: More Scalable Computation via Efficient Approximation

Kim, Young‐Ju; Gu, Chong

doi:10.1046/j.1369-7412.2003.05316.x

Cited by 247 publications

(206 citation statements)

References 28 publications

Supporting

Mentioning

205

Contrasting

Order By: Relevance

“…Various error distributions were applied to the GAMs and, although the resulting fits were comparable, models using transformed cover data with a normal (Gaussian) error distribution, identity link and a gamma value of 1.4 (Kim and Gu 2004) gave the most reliable and robust outcomes. Several other error distributions, for instance a beta distribution, caused an estimate bias when a series included many sequential zero values (BCM, MA, SP and CCA).…”

Section: Univariate Analysesmentioning

confidence: 99%

40 Years of benthic community change on the Caribbean reefs of Curaçao and Bonaire: the rise of slimy cyanobacterial mats

et al. 2017

View full text Add to dashboard Cite

Over the past decades numerous studies have reported declines in stony corals and, in many cases, phase shifts to fleshy macroalgae. However, long-term studies documenting changes in other benthic reef organisms are scarce. Here, we studied changes in cover of corals, algal turfs, benthic cyanobacterial mats, macroalgae, sponges and crustose coralline algae at four reef sites of the Caribbean islands of Curaçao and Bonaire over a time span of 40 yr. Permanent 9 m 2 quadrats at 10, 20, 30 and 40 m depth were photographed at 3-to 6-yr intervals from 1973 to 2013. The temporal and spatial dynamics in the six dominant benthic groups were assessed based on image point-analysis. Our results show consistent patterns of benthic community change with a decrease in the cover of calcifying organisms across all sites and depths from 32.6 (1973) to 9.2% (2013) for corals and from 6.4 to 1% for crustose coralline algae. Initially, coral cover was replaced by algal turfs increasing from 24.5 (1973) to 38% around the early 1990s. Fleshy macroalgae, still absent in 1973, also proliferated covering 12% of the substratum approximately 20 yr later. However, these new dominants largely declined in abundance from 2002 to 2013 (11 and 2%, respectively), marking the rise of benthic cyanobacterial mats. Cyanobacterial mats became the most dominant benthic component increasing from a mere 7.1 (2002) to 22.2% (2013). The observed increase was paralleled by a small but significant increase in sponge cover (0.5 to 2.3%). Strikingly, this pattern of degradation and phase change occurred over the reef slope down to mesophotic depths of 40 m. These findings suggest that reefs dominated by algae may be less stable than previously thought and that the next phase may be the dominance of slimy cyanobacterial mats with some sponges.

show abstract

Section: Univariate Analysesmentioning

confidence: 99%

40 Years of benthic community change on the Caribbean reefs of Curaçao and Bonaire: the rise of slimy cyanobacterial mats

et al. 2017

View full text Add to dashboard Cite

show abstract

“…The arbitrary selection needs to more research and means that if a satisfactory theory for modelling with smooth function is to be extended, the number of knots must be shown. The smoothing parameter, λ, controls the specification between model fit and model smoothness (Kim and Gu, 2004). If λ = 0, means that there is no penalty and if λ→∞, means that something like a straight line is going to estimate f (Wood and Augustin, 2002;Eilers and Marx, 2010).…”

Section: Stationary Datamentioning

confidence: 99%

Univariate Generalized Additive Models for Simulated Stationary and Non-Stationary Generalized Pareto Distribution

Behzadi¹,

Adam²,

Fitrianto³

2017

Journal of Mathematics and Statistics

View full text Add to dashboard Cite

Generalized additive models as a predictor in regression approaches, are made up over cubic spline basis and penalized regression splines. Despite of linear predictor in GLM, generalized additive models use a sum of smooth functions of covariates as a predictor. The data which are used in this study have generalized Pareto distribution and have been simulated by inversion method. The data are generated in two types, the stationary case and the non-stationary case. The method of root mean square of errors as a method of measurement is used for comparison between power of predictions which are based on penalized regression splines as a method in univariate generalized additive models and linear regression based on maximum likelihood estimation. The finding of this research illustrates that the amount of accuracy of estimation of parameter of location in UGAM approach as an alternative promising of modelling through each specialized GPD's models, has less RMSE in compare with MLE.

show abstract

“…Quantile-quantile plots and histograms of residuals were examined to ensure that the distributional assumption was suitable, and plots of linear predictors against residuals revealed variance to be approximately constant for all models. In all cases, the value of γ used in the GCV scores was set at 1.4 to avoid over-fitting (Kim & Gu 2004, Wood 2006) and a gamma distribution was used with a log link function.…”

Section: Functional Responsementioning

confidence: 99%

Functional response of fishers in the Isle of Man scallop fishery

Murray

Hinz

Kaiser

2011

Mar. Ecol. Prog. Ser.

View full text Add to dashboard Cite

To implement effective fisheries management, it is important to understand the variables influencing the distribution and intensity of fishing effort. The functional response of consumers to the availability of prey determines their impact on prey populations. The relationship between predator and prey observed in nature also applies to fishers and the populations they target. The present study focuses on the behaviour of a scallop dredging fleet fishing for Pecten maximus around the Isle of Man during a single fishing season. The functional response was investigated by examining the relationships between catches and fishing effort, scallop abundance and other variables. Scallop abundance was depleted rapidly during the first month of fishing. The increased patchiness of scallops towards the end of the season probably reduced their catchability, but fishers were able to maintain catch rates at intermediate abundance levels. The functional response did not conform to a particular type, but there was latent fishing capacity in the fishing fleet even at the highest levels of abundance. Therefore, reducing the number of vessels would not necessarily reduce fishing mortality unless combined with a reduction in the fishing power of individual vessels.

show abstract

Smoothing Spline Gaussian Regression: More Scalable Computation via Efficient Approximation

Cited by 247 publications

References 28 publications

40 Years of benthic community change on the Caribbean reefs of Curaçao and Bonaire: the rise of slimy cyanobacterial mats

40 Years of benthic community change on the Caribbean reefs of Curaçao and Bonaire: the rise of slimy cyanobacterial mats

Univariate Generalized Additive Models for Simulated Stationary and Non-Stationary Generalized Pareto Distribution

Functional response of fishers in the Isle of Man scallop fishery

Contact Info

Product

Resources

About