Survey Article from the literature

  • Citation of the article

Kumar, Vaibhav, and M. L. Garg. “Predictive analytics: a review of trends and techniques.” International Journal of Computer Applications 182.1 (2018): 31-37.

  • In two or three sentences, discuss the goals of this article.

One goal of this article is to provide a broad picture of today’s applications of predictive analytics. Another one is to provide the core structure and process of predictive analytics in the form of steps that are easy to follow and comprehend.

  • What is /are the general task(s) that all the methods address in the article?

The core task the methods address is the accurate forecasting of events using predictive analytics. This is usually done by creating some model in a myriad of methods.

  • What are some of the leading methods introduced and discussed in this paper?

Some of the leading methods entail gradient boost models, artificial neural networks, and support vector machines. Gradient boost models simulate several decision trees and then try to make one decision tree that’s the “average” of all the others. Artificial neural networks, the leading method I’m interested in, is an imitation of the human brain, in that there’s what’s called neurons that altogether take input and produce output in a mathematical fashion. Support vector machines create plots and represent samples as dots on the plot in order to define categories. Ideally, there’s a region where samples of the same category congregate, and that’s what’s called a support vector.

  • What is the context of the discussion of these methods? For instance, are there demonstrations in light of some task that they address?

The context lies in the sub-purposes of the methods under the guise of predictive analytics. For example, gradient boost models and support vector machines are mainly used for classification purposes and artificial neural networks have multiple purposes including regression and classification. In general, banks use predictive analytics in order to find fraudulent customers and retail uses it in order to predict customer affinities towards certain products.

  • How can some of these methods be used in your own work? What purpose would the method serve in your work?

I can use an artificial neural network to create projections based on previous data. For example, if I were to work on stock market prediction, I would take past financial data and output projections. The method serves a purpose of structure in the implementation.

First article from the literature

  • Citation of the article

Fung, Daryl LX, et al. “Predictive analytics of COVID-19 with neural networks.” 2021 International Joint Conference on Neural Networks (IJCNN). IEEE, 2021.

  • What is the goal of this article?

The goal of the article is to present a solution for predictive analytics of COVID-19 with neural networks.

  • In about 100 words, please summarize the main goal of the article.

This solution is meant to reduce costs of obtaining data by making more effective algorithms that could work with a fraction of the data needed in already existing neural network solutions. They do this by using an autoencoder, which is an artificial neural network that takes in a representation of the data and puts out a cleaner version of the input. An autoencoder has many health-releated uses beyond COVID, one being associations between micro-RNAs and diseases. Another one is Parkinson’s disease. They also used few-shot learning, which allows for less samples being needed to train the model. This is to overcome the limited availability of data.

Second article from the literature

  • Citation of the article

Deepika, Kumari, and S. Seema. “Predictive analytics to prevent and control chronic diseases.” 2016 2nd International Conference on Applied and Theoretical Computing and Communication Technology (iCATccT). IEEE, 2016.

  • What is the goal of this article?

The goal is to provide a solution for predictive analytics of chronic diseases.

  • In about 100 words, please summarize the main goal of the article.

In this case, the solution relates to heart disease and diabetes, both of them being chronic diseases. It uses historical health records to create projections of those chronic diseases. To do that, they use Naive Bayes, decision trees, support vector machines, and artificial neural networks. Naive Bayes is where basic probability calculations are used for the predictions. In this context, Naive Bayes takes in certain conditions and gives a probability of a chronic disease occurring. Decision trees are basically direct classification models, which in this context guess if a chronic disease occurs based on conditions the nodes represent in the tree. Support vector machines organize inputs in a plot so that they can be visually represented and set up to try to determine a relationship with common chronic diseases in a certain region, or support vector. Artificial neural networks are imitations of the human brain, in that there’s what’s called neurons, or nodes, that altogether take input and produce output through mathematical calculations.