From data mining to decision support systems…
Extracting knowledge from data has been a huge topic of discussion in recent years, and has attracted the whole information technology community in the world. The dramatic growth of the data available online and the advanced data storage technologies has made data mining a required task, to identify the knowledge hidden in the data and for gaining insight to drive decision making.
Data mining is widely used in several domains, we will discuss briefly in this story its main applications and its new trends, and why such a task interest the medical community? Moreover, how can it help in the creation of medical aid diagnosis systems?
The first application that can make you realize the interest of such task in our modern life is, the use of data mining systems in supermarkets, the idea is to manage the customer’s data and predict future action based on past actions. Such information is crucial for supermarkets because it helps them to change their layouts accordingly and it makes sense to keep the targeted products close together.
Another interesting application is education, where advanced data mining tools can discover the most effective way to teach students. It can help also to adapt the content of courses based on their skills.
In the financial field, data analysis plays an important role in allowing banks to predict customers behaviour and propose relevant services and products accordingly.
Finally, the analysis of healthcare data can help greatly discover the relationships between diseases, for example, it can inform about the effectiveness of treatments and identify new drugs, or ensure that patients receive appropriate, timely care. It can also predict the number of people falling victim to every pathology and inform the appropriate institutions how can they reduce health costs too.
Data mining process needs several techniques to extract information from the data, in practice, its main high-level goals tend to be Prediction and Description. While the first one involves predicting unknown patterns based on some variables, the second one focuses on finding human-interpretable patterns hidden in the data. Although the boundaries between Prediction and Description are not sharp (some of the predictive models can be descriptive, to the degree that they are understandable, and vice versa) [1]. The ideal case is to find a model that combines perfectly between Description and Prediction goals. Classification, Regression and Clustering are the widely applied by data miners (data scientists).
Classification is a learning task that classifies the data into one of several predefined classes, Regression task has the same learning goal, although, it maps a data item to a real-valued prediction variable. Finally, clustering is a common descriptive task that aims to identify a finite set of categories (clusters) describing the data.
To be continued …
Public: beginners.
