Article: Eagle view: an abstract evaluation of machine learning algorithms based on data properties Journal: International Journal of Computational Intelligence Studies (IJCISTUDIES) 2022 Vol.11 No.1 pp.36 - 52 Abstract: Data can be generated from almost any type of information. Experimental data analysis (EDA) and feature engineering for machine learning models necessitate a thorough understanding of the different types of data. Algorithms that interpret and recall future data details use machine learning (ML) data. The majority of the data can be found on the internet. In terms of machine learning, the majority of the data can be grouped into four categories: numerical data, category data, time-series data, and text. Supervised learning, a collection of unproven learning algorithms, is the subject of this research. Regression models, random forests, logistical regressions, support vector machines, decision trees, neural networks, naive Bayes, t-distributed stochastic neighbourhood embedding (t-SNE), k-means clustering, principal component analysis (PCA), interim variance (TD), Q-learning, and others are among the most recent machine learning approaches. The paper observed the following steps: first is to concentrate on investigating and debating the issues with machine learning, as well as possible solutions; then investigate the conflict time and learning machine release effects on various data types after that; finally, define data types based on what was learned. Inderscience Publishers - linking academia, business and industry through research

Title: Eagle view: an abstract evaluation of machine learning algorithms based on data properties

Authors: Dhairya Vyas; Viral V. Kapadia

Addresses: Computer Science and Engineering Department, The Maharaja Sayajirao University of Baroda, Vadodara, Gujarat, India ' Computer Science and Engineering Department, The Maharaja Sayajirao University of Baroda, Vadodara, Gujarat, India

Abstract: Data can be generated from almost any type of information. Experimental data analysis (EDA) and feature engineering for machine learning models necessitate a thorough understanding of the different types of data. Algorithms that interpret and recall future data details use machine learning (ML) data. The majority of the data can be found on the internet. In terms of machine learning, the majority of the data can be grouped into four categories: numerical data, category data, time-series data, and text. Supervised learning, a collection of unproven learning algorithms, is the subject of this research. Regression models, random forests, logistical regressions, support vector machines, decision trees, neural networks, naive Bayes, t-distributed stochastic neighbourhood embedding (t-SNE), k-means clustering, principal component analysis (PCA), interim variance (TD), Q-learning, and others are among the most recent machine learning approaches. The paper observed the following steps: first is to concentrate on investigating and debating the issues with machine learning, as well as possible solutions; then investigate the conflict time and learning machine release effects on various data types after that; finally, define data types based on what was learned.

Keywords: data types; supervised; unsupervised; reinforcement; outliers; time complexity.

DOI: 10.1504/IJCISTUDIES.2022.123340

International Journal of Computational Intelligence Studies, 2022 Vol.11 No.1, pp.36 - 52

Received: 02 May 2021
Accepted: 08 May 2021
Published online: 10 Jun 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: Eagle view: an abstract evaluation of machine learning algorithms based on data properties

Keep up-to-date