Predictive modeling is a process that uses past data and other attributes to predict outcomes with the data model. Now with this definition think what can or cannot be predictive modeling. I have taken examples to clear your view.
Take the first example of the recommendation of the movie on amazon prime. A recommendation can be done on the basis of past movies watched by a person or on the basis of the same age or gender have watched the movies. This is done by analyzing past data and predicting future forecasts.
The second example is of a factor responsible for sales reduction. This can be found out by analyzing past data, not need to predict the future.
The third example is to predict the default rate for each customer. The default rate for each customer is based on their past data, income, and other attributes.
Another example is viewing website traffic using google analytics. In this, no past data is involved and no need to predict forecasts.
The first and third examples are Predictive modeling. The second and fourth are not predictive modeling. Now you might have a clear understanding of what is predictive modeling or not.
Types of predictive modeling
Supervised learning- In supervised learning, Data is given to predict the target variable. Depending on the nature of the target variable whether is continuous in nature or not supervised learning is divided into two types classification and regression.
consider an example of a classification problem whether a person in titanic will survived or not. It can be predicted with the help of past data or on the basis of gender or age. In this, the target variable is not continuous in nature.
Now, consider an example of a regression problem house price prediction. Price of the house can be predicted on the basis of past data, locality, size, and other attributes. In this, the target variable is of continuous nature.
Unsupervised learning- In unsupervised learning, the task of machine is to group unsorted information according to similarities, patterns, without any prior training of data. A most common example of unsupervised learning is Google news clustering.
Unsupervised learning is dived into two categories.
Clustering -In this type of unsupervised learning, the task is to divide the population or data points into a number of groups such that data points in the same groups are more similar to other data points in the same group than those in other groups.
Association- This algorithm is used to identify new and interesting insights between different objects in a set, frequent pattern in transactional data, or any sort of relational database. An example of this is the market basket analysis.
Comments