Most Fridays, I will answer a question or questions from Principles of Marketing students about marketing. This week's question: What is the best way to predict future trends in customer wants and needs? There is no one way or best way to predict. If there were, then I would teach that method or write that formula on the board. Instead, many forecasting techniques exist.
Broadly, regression offers prediction from a statical approach. This technique allows the manager to understand that these independent variables predict a dependent variable. Regression comes in many flavors such as logic, profit, weighted least squares, ordinary least squares, dummy, classification, etc.
Averages offer another approach to forecast. A simple average (i.e., add two numbers and then divide by two) provides some insight. This approach, however, is sensitive to swings in data. A moving average (i.e., add three numbers and then divide by three) does a better job of handling swings in the data. Alternatively, an instantaneous average, or first derivative, could be advantageous when working with a lot of data that appears within a bounded region such as sales volume of milk.
If the data is particularly lumpy such as stock price movement, then a formula that incorporates the log or exponential functional would be necessary.
If the data is distributed in a binary fashion, then a quadratic equation would be required.
Finally, some forecasters develop simulations such as Monte Carlo to predict future trends. These simulations require a lot of data as well as an experienced modeller.
Given the variety of forecasting techniques, you should choose the technique that best fits your data and your managerial objective.