Decision makers or knowledge workers are looking for information and knowledge at various kinds for making strategic or even mundane decisions. The process of extraction of information and knowledge from data is called data mining. This knowledge and information can be represented in various formats, this could be as simple as mean, median, mode, count or it could be in a graphical format such as histogram, line, pie chart, trend lines or moving averages. Advanced techniques like learning models, business optimizations are next level of knowledge requirment for the organization.
Even a using a tool as simple as a spreadsheet will be extremely helpful in providing a mental representation of business situation. Most commonly used statistical techniques can be implemented in a spreadsheet as simple as MS EXCEL.
There are some very important techniques to business intelligence analysis. Most importantly defining the objective and performance indicators, these are metrics that are used to estimate performance of an object (entity). The next is developing mathematical relationships between variables and metrics through finding patterns. The last is What-If analysis is by determining variations in the output metric by changing the input variables.
The advantage of using mathematical models is beyond increasing performance and ROI. It helps knowledge workers in deeper analysis of the business and underlying product/domain. This will increase awareness in the company, knowledge transfer within the company, and higher desire to learn better things. It encourages intellectual thinking within the company and promote people with good analytical skills who can offer great value to the company.\
There are many techniques like regression and classification, which are some of the popular mathematical models,however predictive analytics are not limited to these methods.
Regression: Linear Regression, kNN, CART, Neural Net
Classification: Logistic Regression, Bayesian Methods, Discriminant Analysis, Neural Net, kNN, CART.
There are some limitations and advantages of each of the methods. The right model and right mathematical technique to be choosen for each problem. The underlining business value that needs to be increased with each of the techniques.
Best techniques are formulates after and testing and evaluating each approach and measuring the impact of success or performance. All mathematical models use simple statistical techniques, however the value is in mapping the business problem into a mathematical problem, this requires some intellectual talent.
Models developed can be extremely useful in business critical process like sales, marketing and product.
More on such topics will be published on http://www.praveenkodur.com/blog/
