Scope of Data Mining

Main goal of data mining is to Extract the Hidden Information or Pattern from Large Dataset and Prediction based on Pattern of Data.

Hidden Information may be Hidden Rules of behavior (Association Rules), Hidden Clusters of Data, the information that can not be extracted using Queries and Reports. Data Mining tools predicts the future behavior based on past behavior. (Training Data and Testing Data). Mostly you can use data mining for Classification and Prediction purpose.

Classification : Classification is used to predict class labels (Categorical) it may be One Class, Two (Binary) Class or Multi Class. Mostly One Class Classification (Unary Classification) is used when you have Imbalanced data. Means You have Training Data Set in which most of Instances are from Single Class, testing data may contains mixed data. Binary Classification is mostly used classification technique that classify the data into Positive or Negative Class. Like based on cancer dataset you can classify whether the instance is Risky or Normal. In Multi-class Classification there are more than two categories of Labels. Like News Classification. Is Social Media Posts are Positive, Negative or Neutral for any topic ? One point to note here is Multi Class Classification is not same as Multi-Label Classification. Each type of classification has their own challenges, so training model and classifier selection plays an important role to achieve better result.

Prediction :

Prediction does not predict any category of class, but mostly it used for Predicting Continuous valued Functions. How Much Sensex (Stock Market) will Up or Down ? , What are the Chances of Profit of particular companies?, Voting Shares of Any Political Party in Upcoming Election and these type of analysis are called Prediction. For Prediction we have to generate prediction model and then apply algorithm of prediction to Predict the future data.