Evaluating the Effect of Dataset Size on Predictive Model Using Supervised Learning Technique

Ajiboye, Adeleke Raheem; Arshah, Ruzaini Abdullah; Qin, Hongwu; Isah-Kebbe, H.

doi:10.15282/ijsecs.1.2015.6.0006

Cited by 56 publications

(39 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…On the whole, the efficiency of the prediction process was found to be approximately 96.365%, indicating that there is a slight deviation between the predictions and the actual values ( Fig 5 ). This can possibly be rectified by increasing the training sample size as generally, it is inferred that using sufficient data set for predictive model construction can lead to better accuracy [ 66 ].…”

Section: Resultsmentioning

confidence: 99%

An apta-aggregation based machine learning assay for rapid quantification of lysozyme through texture parameters

et al. 2021

View full text Add to dashboard Cite

A novel assay technique that involves quantification of lysozyme (Lys) through machine learning is put forward here. This article reports the tendency of the well- documented Ellington group anti-Lys aptamer, to produce aggregates when exposed to Lys. This property of apta-aggregation has been exploited here to develop an assay that quantifies the Lys using texture and area parameters from a photograph of the elliptical aggregate mass through machine learning. Two assay sets were made for the experimental procedure: one with high Lys concentration between 25–100 mM and another with low concentration between 1–20 mM. The high concentration set had a sample volume of 10 μl while the low concentration set had a higher sample volume of 100 μl, in order to obtain the statistical texture values reliably from the aggregate mass. The platform exhibited an experimental limit of detection of 1 mM and a response time of less than 10 seconds. Further, two potential operating modes for the aptamer were hypothesized for this aggregation property and the more accurate mode among the two was ascertained through bioinformatics studies.

show abstract

Section: Resultsmentioning

confidence: 99%

An apta-aggregation based machine learning assay for rapid quantification of lysozyme through texture parameters

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Dataset size is also known to affect the performance of machine learning models (e.g., Ajiboye et al, 2015; Raudys & Jain, 1991). However, the error rates of machine learning models have also been shown to decrease for sample sizes of > 100 observations (Ajiboye et al, 2015). Similar analyses applied using a range of datasets show that accurate results can be achieved for groups with small sample sizes (Bland et al., 2015).…”

Section: Discussionmentioning

confidence: 99%

Morphological traits of reef corals predict extinction risk but not conservation status

Raja

Lauchstedt

Pandolfi

et al. 2021

Global Ecol. Biogeogr.

View full text Add to dashboard Cite

Aim The current assessment of extinction risk in reef corals by the International Union for Conservation of Nature (IUCN) has been criticized, because coral life‐history traits associated with resilience are not reflected in the conservation status. We aimed to carry out a quantitative assessment of the link between reef coral traits and species extinction risk informed by extinctions of reef corals observed in the fossil record. Location Global. Time period Plio‐Pleistocene and present day. Major taxa studied Scleractinian reef corals. Methods We used morphological traits, phylogenetic information and evidence of extinction during the Plio‐Pleistocene to predict the extinction risk of contemporary reef corals. Our model was trained using 138 Caribbean fossil coral species and an automatic machine learning algorithm. We then used this model to predict the extinction risk of 674 modern coral species. Results Model validation confirmed 77% accuracy in predicting extinction risk of fossil corals. Extinction risk predicted by our model showed a near‐random (57%) match with the IUCN conservation status. Our model also suggested that corals in the Least Concern or Near Threatened categories might be at higher risk of extinction than currently believed. Main conclusions Morphological traits of fossil corals linked to their extinction risk in the Plio‐Pleistocene Caribbean Sea are known to reflect the vulnerability of extant corals. However, the results from our fossil‐calibrated model do not match the IUCN assessment of reef corals, with increased overall extinction risk. This does not necessarily indicate near‐future extinction risk, because fossil extinctions are spread over thousands of years. However, we show the applicability of using fossil data to inform the extinction risk of modern corals and recommend that future assessments of extinction risk of reef corals should consider incorporating the relationship between morphological traits and resilience, calibrated by fossil data, to maximize the utility of the extinction risk assessment.

show abstract

“…We would like to note that the results shown in Section 4.2 are based on the test dataset and showed improved results by significantly reducing random and systematic errors, which indicates that our model is successfully calibrated and could potentially be useful to predict the independent hydrometeorological dataset. The machine learning-based error models can manipulate the training data in such a way that the actual results expected from the untrained dataset can be quite different from the evaluated results using the training dataset [51,52,94]. Therefore, we considered the representation of extreme (>95th and <25th) precipitation values in the training and testing dataset to make sure that it covered the entire range of the dataset.…”

Section: Discussionmentioning

confidence: 99%

Machine Learning-Based Error Modeling to Improve GPM IMERG Precipitation Product over the Brahmaputra River Basin

et al. 2020

View full text Add to dashboard Cite

The Integrated Multisatellite Retrievals for Global Precipitation Measurement (GPM) (IMERG) Level 3 estimates rainfall from passive microwave sensors onboard satellites that are associated with several uncertainty sources such as sensor calibration, retrieval errors, and orographic effects. This study aims to provide a comprehensive investigation of multiple machine learning (ML) techniques (Random Forest, and Neural Networks), to stochastically generate an error-corrected improved IMERG precipitation product at a daily time scale and 0.1°-degree spatial resolution over the Brahmaputra river basin. In this study, we used the operational IMERG-Late Run version 06 product along with several meteorological and land surface parameters (elevation, soil type, land type, soil moisture, and daily maximum and minimum temperature) to produce an improved precipitation product in the Brahmaputra basin. We trained, tested, and optimized ML algorithms using 4 years (from 2015 through 2019) of reference rainfall data derived from the rain gauge. The ML generated precipitation product exhibited improved systematic and random error statistics for the study area, which is a strong indication for using the proposed algorithms in retrieving precipitation across the globe. We conclude that the proposed ML-based ensemble framework has the potential to quantify and correct the error sources for improving and promoting the use of satellite-based precipitation estimates for water resources applications.

show abstract

Evaluating the Effect of Dataset Size on Predictive Model Using Supervised Learning Technique

Cited by 56 publications

References 13 publications

An apta-aggregation based machine learning assay for rapid quantification of lysozyme through texture parameters

An apta-aggregation based machine learning assay for rapid quantification of lysozyme through texture parameters

Morphological traits of reef corals predict extinction risk but not conservation status

Machine Learning-Based Error Modeling to Improve GPM IMERG Precipitation Product over the Brahmaputra River Basin

Contact Info

Product

Resources

About