To verify a theory, for example: To analyze the data for new and unexpected relationships, for example: Data mining is a computerized or a manual search for knowledge from huge historical data without prior assumptions about what can be defined. It is an analytical process to explore and search a huge database to extract useful patterns and relationships and to find the correlation between its elements. Data mining is a new technology that enables the predictive pattern discovery, hypothesis creation and testing, and insight-provoking generation.
- Applied Information Science, Engineering and Technology: Selected Topics from the Field of Production Information Engineering and IT for Manufacturing: ... in Intelligent Engineering and Informatics);
- Experimental Design: Procedures for the Behavioral Sciences.
- Talent, Competitiveness and Migration: The Transatlantic Council on Migration.
Data mining which is also known as Knowledge Discovery in Database KDD is any application which has a capability for extracting hidden knowledge and it is not related to any specific industry. It is considered as one of the top ten information technology aspects that will change the world in the coming years. Data mining process blends between artificial intelligence science, statistics, machine learning and databases. We stand daily in long line at big supermarkets waiting for our turn to pay the value of some of our purposes which we have purchased.
During this period of waiting we hear multiple beeps coming out from many barcode readers. We know that any barcode beep is a transaction, and represented by a purchase record stored in the database. So, hundred thousands of records can be accumulated in the database per day. These records, however, contain important information of purchase process for many items and also the best-selling items.
But how can we benefit from all of this data and how to make it useful in our business? Data mining technology only can answer this question. The problem is not like the past concentrated in the lack of storage space and insufficient data but, in fact, our lack in the experiences that is capable to convert this data to a valuable knowledge.
As we said before, data mining technology is the process of extrapolation in a large volume of data in order to detect influencing factors of a particular behavior such as the causes, conditions, contraindications etc. Data mining is not directed to a specific domain but it has a clear impact in all life aspects. So, we can summarize its importance in information industry and why we use it in the following points:. Data mining techniques capable to answer complex questions in record time, especially the types of questions those has been difficult to find answers to them by using classical statistical techniques.
The discovery of knowledge in a database is a process connected with managers and decision makers who are involved in results implementation. The KDD process consists of seven stages and can be summarized as follows:. Stages 1 through 4 represent data preprocessing procedures i. Fig 1 illustrates a general overview of data mining system architecture. We notice that the data mining is only one step in KDD and it consists of complex data miner applications for knowledge extraction and the resulting knowledge may be stored further in a knowledge base.
Referring to fig 1 we can say that the structure of data mining consists of six components as follows:. Data warehouse, flat files, database, World Wide Web or any other data storage container where the data preprocessing techniques take place. Database or data warehouse server which is dealing with retrieving and capturing data according to the data mining user requests. Knowledge base which is used for storing the extracted knowledge for further evaluation of the resulting patterns.
Data mining engine which consists of data mining applications modules that perform all data mining functions such as classification, summarization, clustering, association rules, prediction, time series analysis, regression and sequence discovery. Pattern evaluation module which communicates with the data mining modules for measuring the interestingness and focusing the search towards interesting patterns. User interface is responsible of communication between the end user and the data mining modules and provides the end user ability to perform his query tasks, perform all exploratory data mining, browsing databases and data warehouses and evaluate their schemas and data structures.
The KDD knowledge output can be accommodated in decision making, query processing, information management and process control. Therefore, according to Jiawei Han and Micheline Kamber " Data mining is considered one of the most important frontiers in database and information systems and one of the most promising interdisciplinary developments in the information technology" p Data preprocessing is considered to be the very serious stage in the data mining and the correct exploration in database should be built on a data that ensures the flow of knowledge.
The databases, as we know, contain groups of a very large amount of data that is collected through certain automated methods that are not completely controlled. So the databases are vulnerable to missing, incorrect, incomplete, inconsistent and noisy data which represent the inputs to analysis processes and therefore the knowledge discovery. An attention should be paid to the quality of data, if not collected and selected carefully may leads to misleading results specifically in the predictive data mining.
The data preprocessing methods such as data selection, data cleaning, data integration and data transformation should be applied to the database to correct errors, remove noisy data and gathering data from various data sources. Descriptive data summarization techniques can be applied first to highlight the data properties and distinguish between noisy, missing, incorrect, outliers and incomplete data. Human data entry contributes to an inaccurate and missing data and it is data cleaning process functions to deal with the data entry errors.
Data integration process is responsible for well designed schema in its tables, attributes and constraints, i. The data transformation process is concerned with data structure i. The ETL software which allows extract, transform and load data to and from a database or a data warehouse play an important role in the data transformation process. The data mining methods can be grouped into two major types: Descriptive data mining such as summarization, clustering, association rules and sequence discovery deals with the general characteristics of data in the database and it depends on the reorganization of data and mining in its depths as if for the extraction of models that allows you to create a simple description of similar entities such as similar customers in sales database and no target is required for such data.
The predictive data mining, on the other hand, is trying to find the best predictions based on the data, such as knowing the best and the preferred product to a specific customer. In brief, this type of data mining depends on the historical data i. Unlike descriptive data mining, the predictive data mining has a target to achieve. The descriptive data mining tasks can be summarized in classification, regression, time series analysis and prediction. In the following paragraph we will highlight some of data mining tasks which include classification, clustering, association rule, sequence discovery, regression, and time series analysis.
The following overview will only list the most prominent examples of clustering algorithms, as there are possibly over published clustering algorithms. Not all provide models for their clusters and can thus not easily be categorized. An overview of algorithms explained in Wikipedia can be found in the list of statistics algorithms. There is no objectively "correct" clustering algorithm, but as it was noted, "clustering is in the eye of the beholder. It should be noted that an algorithm that is designed for one kind of model will generally fail on a data set that contains a radically different kind of model.
Connectivity-based clustering, also known as hierarchical clustering , is based on the core idea of objects being more related to nearby objects than to objects farther away. These algorithms connect "objects" to form "clusters" based on their distance. A cluster can be described largely by the maximum distance needed to connect parts of the cluster. At different distances, different clusters will form, which can be represented using a dendrogram , which explains where the common name "hierarchical clustering" comes from: In a dendrogram, the y-axis marks the distance at which the clusters merge, while the objects are placed along the x-axis such that the clusters don't mix.
Connectivity-based clustering is a whole family of methods that differ by the way distances are computed. Apart from the usual choice of distance functions , the user also needs to decide on the linkage criterion since a cluster consists of multiple objects, there are multiple candidates to compute the distance to use. Furthermore, hierarchical clustering can be agglomerative starting with single elements and aggregating them into clusters or divisive starting with the complete data set and dividing it into partitions.
These methods will not produce a unique partitioning of the data set, but a hierarchy from which the user still needs to choose appropriate clusters. They are not very robust towards outliers, which will either show up as additional clusters or even cause other clusters to merge known as "chaining phenomenon", in particular with single-linkage clustering. In the data mining community these methods are recognized as a theoretical foundation of cluster analysis, but often considered obsolete [ citation needed ].
They did however provide inspiration for many later methods such as density based clustering.
Cluster analysis - Wikipedia
Single-linkage on Gaussian data. At 35 clusters, the biggest cluster starts fragmenting into smaller parts, while before it was still connected to the second largest due to the single-link effect. Single-linkage on density-based clusters. In centroid-based clustering, clusters are represented by a central vector, which may not necessarily be a member of the data set. When the number of clusters is fixed to k , k -means clustering gives a formal definition as an optimization problem: The optimization problem itself is known to be NP-hard , and thus the common approach is to search only for approximate solutions.
A particularly well known approximate method is Lloyd's algorithm ,  often just referred to as " k-means algorithm " although another algorithm introduced this name. It does however only find a local optimum , and is commonly run multiple times with different random initializations. Most k -means-type algorithms require the number of clusters — k — to be specified in advance, which is considered to be one of the biggest drawbacks of these algorithms. Furthermore, the algorithms prefer clusters of approximately similar size, as they will always assign an object to the nearest centroid.
This often leads to incorrectly cut borders of clusters which is not surprising since the algorithm optimizes cluster centers, not cluster borders. K-means has a number of interesting theoretical properties.
First, it partitions the data space into a structure known as a Voronoi diagram. Second, it is conceptually close to nearest neighbor classification, and as such is popular in machine learning. Third, it can be seen as a variation of model based clustering, and Lloyd's algorithm as a variation of the Expectation-maximization algorithm for this model discussed below. K-means separates data into Voronoi-cells, which assumes equal-sized clusters not adequate here.
The clustering model most closely related to statistics is based on distribution models. Clusters can then easily be defined as objects belonging most likely to the same distribution. A convenient property of this approach is that this closely resembles the way artificial data sets are generated: While the theoretical foundation of these methods is excellent, they suffer from one key problem known as overfitting , unless constraints are put on the model complexity.
A more complex model will usually be able to explain the data better, which makes choosing the appropriate model complexity inherently difficult. One prominent method is known as Gaussian mixture models using the expectation-maximization algorithm. Here, the data set is usually modeled with a fixed to avoid overfitting number of Gaussian distributions that are initialized randomly and whose parameters are iteratively optimized to better fit the data set.
This will converge to a local optimum , so multiple runs may produce different results. In order to obtain a hard clustering, objects are often then assigned to the Gaussian distribution they most likely belong to; for soft clusterings, this is not necessary.
Data Mining - a search for knowledge
Distribution-based clustering produces complex models for clusters that can capture correlation and dependence between attributes. However, these algorithms put an extra burden on the user: On Gaussian-distributed data, EM works well, since it uses Gaussians for modelling clusters. In density-based clustering,  clusters are defined as areas of higher density than the remainder of the data set. Objects in these sparse areas - that are required to separate clusters - are usually considered to be noise and border points.
Similar to linkage based clustering, it is based on connecting points within certain distance thresholds. However, it only connects points that satisfy a density criterion, in the original variant defined as a minimum number of other objects within this radius. A cluster consists of all density-connected objects which can form a cluster of an arbitrary shape, in contrast to many other methods plus all objects that are within these objects' range.
Another interesting property of DBSCAN is that its complexity is fairly low — it requires a linear number of range queries on the database — and that it will discover essentially the same results it is deterministic for core and noise points, but not for border points in each run, therefore there is no need to run it multiple times. On data sets with, for example, overlapping Gaussian distributions — a common use case in artificial data — the cluster borders produced by these algorithms will often look arbitrary, because the cluster density decreases continuously.
On a data set consisting of mixtures of Gaussians, these algorithms are nearly always outperformed by methods such as EM clustering that are able to precisely model this kind of data. Mean-shift is a clustering approach where each object is moved to the densest area in its vicinity, based on kernel density estimation.
Eventually, objects converge to local maxima of density. Similar to k-means clustering, these "density attractors" can serve as representatives for the data set, but mean-shift can detect arbitrary-shaped clusters similar to DBSCAN. Besides that, the applicability of the mean-shift algorithm to multidimensional data is hindered by the unsmooth behaviour of the kernel density estimate, which results in over-fragmentation of cluster tails. In recent years, considerable effort has been put into improving the performance of existing algorithms.
This led to the development of pre-clustering methods such as canopy clustering , which can process huge data sets efficiently, but the resulting "clusters" are merely a rough pre-partitioning of the data set to then analyze the partitions with existing slower methods such as k-means clustering. For high-dimensional data , many of the existing methods fail due to the curse of dimensionality , which renders particular distance functions problematic in high-dimensional spaces.
This led to new clustering algorithms for high-dimensional data that focus on subspace clustering where only some attributes are used, and cluster models include the relevant attributes for the cluster and correlation clustering that also looks for arbitrary rotated "correlated" subspace clusters that can be modeled by giving a correlation of their attributes.
Several different clustering systems based on mutual information have been proposed. Evaluation or "validation" of clustering results is as difficult as the clustering itself. Internal evaluation measures suffer from the problem that they represent functions that themselves can be seen as a clustering objective.
Techniken Des Data Mining, Knowledge Discovery Und SPSS (German, Paperback)
For example, one could cluster the data set by the Silhouette coefficient; except that there is no known efficient algorithm for this. By using such an internal measure for evaluation, one rather compares the similarity of the optimization problems,  and not necessarily how useful the clustering is. External evaluation has similar problems: On the other hand, the labels only reflect one possible partitioning of the data set, which does not imply that there does not exist a different, and maybe even better, clustering. Neither of these approaches can therefore ultimately judge the actual quality of a clustering, but this needs human evaluation,  which is highly subjective.
Nevertheless, such statistics can be quite informative in identifying bad clusterings,  but one should not dismiss subjective human evaluation. When a clustering result is evaluated based on the data that was clustered itself, this is called internal evaluation. These methods usually assign the best score to the algorithm that produces clusters with high similarity within a cluster and low similarity between clusters.
One drawback of using internal criteria in cluster evaluation is that high scores on an internal measure do not necessarily result in effective information retrieval applications. For example, k-means clustering naturally optimizes object distances, and a distance-based internal criterion will likely overrate the resulting clustering. Therefore, the internal evaluation measures are best suited to get some insight into situations where one algorithm performs better than another, but this shall not imply that one algorithm produces more valid results than another.
An algorithm designed for some kind of models has no chance if the data set contains a radically different set of models, or if the evaluation measures a radically different criterion. On a data set with non-convex clusters neither the use of k-means, nor of an evaluation criterion that assumes convexity, is sound. More than a dozen of internal evaluation measures exist, usually based on the intuition that items in the same cluster should be more similar than items in different clusters.
In external evaluation, clustering results are evaluated based on data that was not used for clustering, such as known class labels and external benchmarks. Such benchmarks consist of a set of pre-classified items, and these sets are often created by expert humans.
Thus, the benchmark sets can be thought of as a gold standard for evaluation. However, it has recently been discussed whether this is adequate for real data, or only on synthetic data sets with a factual ground truth, since classes can contain internal structure, the attributes present may not allow separation of clusters or the classes may contain anomalies. A number of measures are adapted from variants used to evaluate classification tasks.
- The Buzzin Guns in the Battle for Light.