Data Mining

The chapter essentially introduces us about the ways to deal with business problems, different DM algotrihtms and their associated tasks, the DM methods, the DM process, other techniques and technologies and few solutions to the business questions that might arise eventually.

We all know that there are several DM algorithms….

Even though there are different Data mining algorithms developed over the years these algorithms typically address only a handful of different data mining tasks.

We will discuss some of these tasks in one line and not in detail as we also have to skim through the rest of the chapter.. An example is the best way to explain any concept. Here it goes.

1. Classification:  For eg: in Product marketing, the classification output attempts to predict/ or tries to answer the questions like will the customer respond to this product offer or not? and accordingly segregates each customer into different set of classes. So the possible classes here are will respond and will not respond.
2. Regression: it usually estimates the numerical value of particular variable specific to each individual. An example regression question would be: will the customer respond to this service or not?
3. Similarity Matching: it identifies similar entities based on the known data.
4. Clustering: it is a process of partitioning a set of data into meaningful sub-classes based on their similarity. These sub-classes are called clusters. To give you Examples of clustering applications well… I can think of marketing, insurance and many more areas like these.
5. Co-occurrence
6. Profiling
7. Link Prediction
8. Data reduction is simply doing away with less significant data and converting the larger dataset to a smaller one for ease of handing and processing it.
9. Casual modeling

Supervised Versus Unsupervised Methods

Supervised methods: lets talk about Supervised DM methods first. If we have a specific purpose in mind for grouping or if we segment our data based on a specific reason it is called supervised DM problem.

EG: Classification, regressions, causal modeling

Contrastly if there is no specific purpose or target being specified for grouping then such techniques are called unsupervised ones.

Eg: Clustering, co-occurrence grouping and profiling are generally unsupervised methods.

...