Outliers Detection in Regression Analysis Using Partial Least Square Approach

Devarakonda, Nagaraju; Shaik, Subhani; Basha, Shaik Althaf Hussain

doi:10.1007/978-3-319-03095-1_14

Cited by 4 publications

(3 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…OD is a key tool in safeguarding data quality and improving the model performance [43]. PLSR has been used for OD by excluding the data points not well described by the model to improve the calibration performance [44][45][46]. It is known that the presence of outliers strongly influences the PLSR method, and the models obtained may not describe the data well and impact overall model effectiveness [47].…”

Section: Discussionmentioning

confidence: 99%

The effect of laser wavelength on soil carbon measurement using laser-induced breakdown spectroscopy

Mohajan,

Huang,

Beier

et al. 2023

Preprint

View full text Add to dashboard Cite

We investigate the effect of laser wavelength on laser-induced breakdown spectroscopy (LIBS) on the measurement of carbon in agricultural soils. Two laser wavelengths, 1064 nm and 532 nm were used to determine soil carbon concentration. No chemical pretreatment, grinding, or pelletization was performed on soil samples to simulate in-field conditions. A multivariate calibration model with outlier filtering and optimized parameters in partial least squared regression (PLSR) was established and validated. The calibration model estimated carbon content in soils with an average prediction error of 4.7% at a laser wavelength of 1064 nm and 2.7% at 532 nm. The limit of detection (LOD) range for 532 nm was 0.34-0.5 w/w%, approximately half of the LOD range for 1064 nm laser wavelength. The improvement in prediction error and LOD of LIBS measurements is attributed to the increase in plasma density achieved at 532 nm.

show abstract

Section: Discussionmentioning

confidence: 99%

The effect of laser wavelength on soil carbon measurement using laser-induced breakdown spectroscopy

Mohajan,

Huang,

Beier

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…It deviates from other objects such that it makes suspicious to be generated by a different mechanism [10]. Subject of outlier is either an elimination of disturbance such as noise reduction [10][11][12][13][14], or an interest of detection such as crime detection [15][16][17][18][19][20][21][22][23]. This paper focuses on the former view.…”

Section: Introductionmentioning

confidence: 99%

Outlier elimination using granular box regression

et al. 2016

View full text Add to dashboard Cite

“…TP was the summation of inorganic, organic, and dissolved forms of phosphorus; and TN was the summation of nitrate, nitrite (NO2 -), ammonia (NH3), and organic nitrogen. An interquartile range (IQR) criteria with a factor of two (i.e., median ± 2*IQR) was used to remove outliers, which can distort the goodness of model fit (Aggarwal and Yu, 2001;Devarakonda et al, 2014). The sample size (n) of the temporal datasets ranged from 16 to 241 among the different sites (Table 1).…”

Section: Datasetsmentioning

confidence: 99%

Relative Linkages of Stream Dissolved Oxygen with the Environmental Drivers across the Gulf Coast of U.S.A

Gebreslase¹

View full text Add to dashboard Cite

Relative Linkages of Stream Dissolved Oxygen with the Environmental Drivers across the Gulf Coast of U.S.A. Aron Gebreslase Dynamics of coastal stream water quality and ecosystem health is influenced by a multitude of interacting environmental drivers. A systematic data analytics approach was employed to determine the relative environmental controls on stream dissolved oxygen (DO) dynamics across the Gulf Coast of U.S.A. Pearson's correlation, principal component, and factor analyses were utilized to examine the interrelations among DO, solar radiation, water temperature, atmospheric pressure, flow rate, nutrients, pH, and specific conductance (a surrogate of salinity) in 36 streams. Relative linkages of DO and the environmental drivers were then reliably estimated by resolving multicollinearity with power-law based partial least squares regression (Nash-Sutcliffe efficiency = 0.58-0.94). Based on the dominant controls of DO, streams were grouped into three regions across the U.S. Gulf Coast. In the northern part of Gulf Coast states (Texas, Louisiana, Mississippi, Alabama, and West Florida), water temperatures had the strongest and dictated control on stream DO. However, in the southern part of Texas and Florida coasts, pH showed the most dominant control on stream DO. Further, streams in between these two regions demonstrated notable controls of multiple drivers (water temperature, stream flow, and specific conductance) on DO. Four dynamic process components adequately described the system data variance in all three regions along the Gulf Coast. For example, the 'climate component' (temperature, solar radiation) in the northern part of Gulf Coast showed 2.7, 3.1, and 3.6 times stronger linkages with stream DO than that of the redox (pH, specific conductance), nutrient (total nitrogen, total phosphorus), and hydro-atmospheric (flow rate, atmospheric pressure) components, respectively. In contrast, in the southern part of Gulf Coast region, the redox components showed 1.6, 2.3, and 2.6 times stronger linkages with stream DO than that of the climate, nutrient, and hydro-atmospheric components, respectively. The identified environmental regimes and estimated linkages of stream DO provide important information into the dominant drivers and dynamic process components of water quality in urban/natural streams across the U.S. Gulf Coast. The knowledge and insights would help coastal managers and stakeholders to achieve a good stream water quality and ecosystem health.

show abstract

Outliers Detection in Regression Analysis Using Partial Least Square Approach

Cited by 4 publications

References 7 publications

The effect of laser wavelength on soil carbon measurement using laser-induced breakdown spectroscopy

The effect of laser wavelength on soil carbon measurement using laser-induced breakdown spectroscopy

Outlier elimination using granular box regression

Relative Linkages of Stream Dissolved Oxygen with the Environmental Drivers across the Gulf Coast of U.S.A

Contact Info

Product

Resources

About