Five Killer Techniques for Data Mining

Feb 11, 2013
Scott Raspa

If the idea of “big data” intimidates you or if the concept of “data mining” sounds like intense labor to you, you’re not alone. Although powerful information can be gathered from infinite sources at any given time, it’s only natural that the majority of businesses have no idea where to begin in collecting relevant data.

Because of this, data mining softwares and companies have emerged in order to assist corporations and governmental entities in collecting and organizing big data in such a way that it is able to help achieve specific objectives. This article will explore the various data mining techniques in use today, and how your organization can use them to accomplish real-world tasks.


Five Killer Techniques for Data Mining - AssociationAssociation is one of the better known data mining techniques. Association strives to discover patterns in data which are based upon relationships between items in the same transaction. Because of its nature, association is sometimes referred to as “relation technique”.

This method of data mining is utilized within market based analysis in order to identify a set, or sets of products that consumers often purchase at the same time. Of the various data mining techniques, association is the most useful for retailers.

As an example, a store may discover that its customers frequently buy potato chips and beer at the same time. By placing a display of chips near the beer cooler, more sales will be encouraged and customers will feel more satisfied as they are able to save time.


Five Killer Techniques for Data Mining - ClissificationClassification is one of the classic data mining techniques which is based upon machine learning. In this method, each item in a set of data is classified into a predefined set of classes or groups.

This goal is achieved through the utilization of mathematical techniques ( ie: decision trees, linear programming and statistical equations) which can be used to program and develop software programs which are then able to classify the data items into groups.

Consider a classification application that is programmed to categorize “given the records of old leads which were successfully converted into sales, which current leads are most likely to buy X product”.

From here, the software can classify sales leads into those who are most likely to make a buying decision, and those that are most likely dead ends. In this way, your sales team can put more focus on those who are likely to close the deal.


ClusteringClustering is utilized to create meaningful and/or useful groups or “clusters” of data that has similar characteristics. The automated technique of clustering both defines the classes, and puts data into each class. This differs from classification in that the classification technique simply assigns data to predefined classes.

This concept can be compared to a library. Libraries house a wide range of books in numerous topics. In order to simplify the process of finding books, a library will “cluster” certain genres of books together in a logical format.

Similarly, the clustering technique of data mining will group similar sets of data together in order to make it easier for end users to find what they are looking for. This is especially useful in litigation and eDiscovery when certain types of data must be retrieved quickly.


PredictionAs the name suggests, prediction is used to make educated guesses about the future. This is accomplished through the utilization of data mining softwares designed to unveil relationships between independent variables, as well as relationships between dependent and independent variables.

As an example, the prediction technique could be used in sales in order to predict profits. If the sale is considered as an independent variable and the profit is the dependent variable, historic sale and profit information could be create to draw a fitted regression curve to accurately guess trends for profits in the future.

Sequential Patterns

Sequential PatternsThe sequential pattern method is another of the data mining techniques that is extremely useful for analyzing behavior in various industries. This analysis strategy works to identify similar patterns, regular events, or trends found within transaction data throughout a specific time period. Sequential patterns in historical transaction data can provide businesses with a clear picture of certain items or sets of items that consumers tend to buy at different times throughout the week or year.

As an example, a grocery store may use sequential patterns to learn that soda and alcoholic beverages are most frequently purchased on Thursdays as customers prepare for the weekend. The store can then make certain to charge full price for these items on Thursdays in order to maximize profits. The end result of these data mining techniques is an all-encompassing “big picture” of what is going on within your business.

Data mining techniques can improve customer relations, boost revenue, make informed guesses about profits, simplify the task of finding relevant information when you need it, and more! It’s time to start embracing big data and all that it can accomplish for your organization.

Want to take the next step with data mining?

IKANOW’s Infinit.e open analytics platform is here to help. With this valuable and flexible big data tool, you can organize and enrich data sources to generate actionable intelligence. We also have free eBooks, whitepapers and videos to help you dive into the powerful and growing field of big data analytics.

Learn more:

Visit the Resource Center

No comments

Leave a Reply

Your email address will not be published. Required fields are marked *