Research methods are employed to broaden the horizons of knowledge, research methods applied to forecast future behavior of any variable is called forecast based research. This type of research is applied to obtain sales forecasts, manpower forecasts, availability of resources, economic forecasts, in demographic studies like population growth, employment levels, income distribution, political forecasting, weather forecasting, seismic forecasting and in many more areas.
Depending upon the nature of the problem, researcher can apply qualitative or quantitative research; sometimes in complex problems both methods can be applied. Qualitative methods include Delphi method, focus groups, opinions of involved people, cookie’s method, analogy forecasting, scenario building, technology forecasting. Quantitative methods include causal methods and time series methods. Causal methods include leading indicators, regression and econometric models and input output analysis. Time series method includes methods like trend projection, trend fitting, exponential smoothing, autoregressive model and Box-Jenkins model. Other advanced methods like artificial intelligence method, simulation and probabilistic methods have also emerged.
The forecasting process comprises determining forecasting purpose, time period, whether short term, medium term or long term, selection of method of forecasting, collection of data and finally analysis and interpretation of data. Selection of method is generally based on three major considerations are those are of accuracy, data requirement and time and cost constraints. The data can be time series, longitudinal or cross-sectional data and methods would vary as per data type.
Forecasts are used as a planning and decision-making tool, entities have to work in changing and risk based environment and future performance is based on the ability to correctly gauge change in its various dimensions and degrees. The future arises from the past and present hence projection and prediction techniques when correctly applied can lead to accurate forecasts, the occurrence of totally unexpected future phenomenon cannot be ruled out, but by planning for the expected uncertainty can definitely be reduced.