Registered Data


[02198] Application of Machine Learning Algorithms for Air Quality Classification

  • Session Date & Time : 3E (Aug.23, 17:40-19:20)
  • Type : Contributed Talk
  • Abstract : Machine learning algorithms build a model based on sample data in order to make predictions or decisions without being explicitly programmed to do so. This research is focused on exploring air quality data using some classification algorithms: random forest, naive Bayes, and support vector machine. We use air quality data in Jakarta, Indonesia, from IQAir World Air Quality Report. As a result, the random forest algorithm gives the best performance with an accuracy of 99.3%.
  • Classification : 62-08, 62R07, 62-11
  • Author(s) :
    • Hasih Pratiwi (Universitas Sebelas Maret)
    • Qonita Indriyani (Universitas Sebelas Maret)
    • Sri Sulistijowati Handajani (Universitas Sebelas Maret)

[02549] Copula for Markov Chain Model with Binomial Time Series Data

  • Session Date & Time : 3E (Aug.23, 17:40-19:20)
  • Type : Contributed Talk
  • Abstract : Markov Chain Model can be constructed by the concept of time dependency and approached by the Copula. Markov Chain Model using Clayton and Joe Copula is proposed to determine the 3-sigma control limit in statistical control process for time series binomial data. In this paper, Gumbel and Frank Copula use in determining the 3-sigma control limits. We conduct simulations to see the performance of the developed methods and analyze the number of defective pieces in the jewelry manufacturing process.
  • Classification : 62-08, 62A99
  • Author(s) :
    • Pepi Novianti (Universitas Gadjah Mada, University of Bengkulu)
    • Gunardi Gunardi (Universitas Gadjah Mada)
    • Dedi Rosadi (Universitas Gadjah Mada)

[02191] Analysis Seismic Data in Sumatra Using Robust Sparse K-Means Clustering

  • Session Date & Time : 3E (Aug.23, 17:40-19:20)
  • Type : Contributed Talk
  • Abstract : K-means algorithm is considered to be the most important unsupervised machine learning method in clustering. It works intimately on complete and clear data but cannot handle outliers. Therefore, robust statistical algorithms are required to deal with it. This paper presents robust sparse k-means algorithm to show clustering of seismic data in Sumatra. Clustering results are displayed graphically for two, three and four clusters to see the zones formed based on the grouping results.
  • Classification : 62-11, 62-08
  • Author(s) :
    • Ulfasari Rafflesia (Univestitas Gadjah Mada, Yogyakarta)
    • Dedi Rosadi (Univestitas Gadjah Mada, Yogyakarta)
    • Devni Prima Sari (Universitas Negeri Padang)

[01056] Statistical methodology for functional meta-analysis of sex-based disparities in neurological diseases

  • Session Date & Time : 3E (Aug.23, 17:40-19:20)
  • Type : Industrial Contributed Talk
  • Abstract : Sex-based differences in diverse health scenarios and diseases have been acknowledged for many years but still not thorough-fully analysed. We propose a statistical methodology combining transcriptomics data from different spurces which allows to unveil those disparities at the level of differentially expressed genes and differentially enriched functions. The methodology uses linear models, meta-analysing the results through the logFC, and has been successfully applied to various diseases such as Multiple Sclerosis, Alzheimer’s and Parkinson’s diseases.
  • Classification : 62-xx, 62Pxx, 92-04, 92-08, 92-10
  • Author(s) :
    • Marta R. Hidalgo (CIPF)
    • Francisco Garcia-Garcia (CIPF)
    • Borja Gomez-Cabañes (CIPF)
    • Carla Perpiña-Clerigues (CIPF)
    • Irene Soler-Saez (CIPF)
    • Fernando Gordillo-González (CIPF)
    • Gonzalo Anton-Bernat (Universitat de València)
    • Adolfo López-Cerdan (BioBam )
    • Rubén Grillo-Risco (CIPF)
    • Jose Francisco Català-Senent (INCLIVA)

[00076] The effect of Anderson acceleration on nonlinear solver convergence order

  • Session Date & Time : 3E (Aug.23, 17:40-19:20)
  • Type : Contributed Talk
  • Abstract : We consider Anderson acceleration (AA) applied to superlinearly and sublinearly converging nonlinear solvers. Recent work has only solved how AA affects linearly converging solvers. We prove $m=1$ AA changes order r to $\frac{r+1}{2}$, and then generalize to higher m. This holds for superlinear and locally for sublinear convergence, thus improving sublinear convergence. The order for AA-Newton is also determined. Tests for nonlinear Helmholtz, Navier-Stokes, and regularized Bingham illustrate the theory.
  • Classification : 65Bxx, 76Dxx
  • Author(s) :
    • Leo Rebholz (Clemson University)