Unit 2: Regression-based forecasting
Business Forecasting notes · PTU syllabus (MBA 964-18)
On this page
Unit summary
Regression links demand to its drivers, so managers can forecast and test what-if scenarios. This unit covers marketing research techniques — consumer surveys and market experiments — statistical estimation, variable identification, time series and cross-sectional data, regression model specification and interpretation, and forecasting with regression models.
After this unit you can
- Use consumer surveys and market experiments to estimate demand
- Identify variables and choose data types
- Specify and interpret regression models
- Forecast with regression models
PTU syllabus topics
- Marketing research techniques
- consumer surveys and market experiments
- statistical estimation
- variable identification
- time series and cross-sectional data
- regression model specification and interpretation
- forecasting with regression models
- 1
Identify variables
- 2
Collect time series or cross-section data
- 3
Specify the model
- 4
Estimate coefficients
- 5
Test assumptions
Multicollinearity, autocorrelation
- 6
Forecast and check accuracy
Topic 1
Marketing research techniques for demand estimation
- Consumer surveys: complete enumeration, sample surveys, end-use method; ask buying intentions, willingness to pay.
- Market experiments: vary price, advertising or packaging in test markets and observe sales; laboratory experiments (simulated stores, conjoint analysis).
- Limitations: stated intentions differ from behaviour; experiments are costly and may be influenced by competitors.
Topic 2
Statistical estimation and variable identification
- Dependent variable: quantity demanded or sales.
- Independent variables: own price, competitors' prices, income, advertising, population, season, distribution, credit availability.
- Choosing variables: economic theory, managerial knowledge, data availability, correlation analysis; avoid highly correlated predictors.
Topic 3
Time series and cross-sectional data
Observations
One unit over many periods
Many units at one point in time
Example
Monthly sales of a brand for 5 years
Sales across 100 cities in 2025
Issues
Autocorrelation, trend, seasonality
Heteroscedasticity, unit differences
Use
Forecasting over time
Effects of drivers across markets
- Panel data combines both — many units over many periods.
Topic 4
Regression model specification and interpretation
Linear
Q = a + b1 P + b2 I + b3 A + e
Log-linear (constant elasticity)
ln Q = a + b1 ln P + b2 ln I — coefficients are elasticities
Interpretation
b1: change in Q for a one-unit change in P, holding others constant
- Evaluate the model: signs as expected, t-tests (p-values) for coefficients, R² and adjusted R², F-test, standard error of estimate, residual checks (multicollinearity, heteroscedasticity, autocorrelation — Durbin–Watson).
Example
ln Q = 4.2 − 1.3 ln P + 0.8 ln I: price elasticity −1.3 (elastic), income elasticity 0.8 (normal necessity).
Topic 5
Forecasting with regression models
- Steps: forecast or assume values of independent variables, plug into the equation, compute the point forecast and a prediction interval (forecast ± t × standard error of forecast).
- Caution: errors in forecasting the drivers, structural changes, extrapolation beyond the data range.
Example
Sales = 120 + 2.5 × advertising (₹ lakh) − 4 × price (₹). With advertising ₹20 lakh and price ₹15, forecast = 120 + 50 − 60 = 110 thousand units.
Key terms
- Market experiment
- Testing demand responses in real or simulated markets
- Cross-sectional data
- Observations of many units at one time
- Log-linear model
- Regression in logs whose coefficients are elasticities
- Prediction interval
- Range likely to contain the actual value
- Standard error of estimate
- Average deviation of actual from fitted values
Quick revision
- Surveys, experiments; limitations.
- Dependent and independent variables; selection.
- Time series, cross-sectional, panel data.
- Linear and log-linear specification; tests and diagnostics.
- Regression forecasts and prediction intervals.
Important exam questions
Practice questions written to the PTU exam pattern for this unit's syllabus: short answers (Section A style) and long answers (Sections B and C style).
Short-answer questions
- Q1.State two limitations of consumer surveys.
- Q2.What is a market experiment?
- Q3.Distinguish time series and cross-sectional data.
- Q4.Why are log-linear coefficients elasticities?
- Q5.Name three tests used to evaluate a regression model.
- Q6.What is a prediction interval?
Long-answer questions
- Q1.Explain consumer surveys and market experiments for demand estimation.
- Q2.Discuss variable identification and data types in demand estimation.
- Q3.Explain the specification and interpretation of regression models for demand.
- Q4.Explain forecasting with regression models with an example.
Stuck on this unit?
Message SBS on WhatsApp for help with Business Forecasting, or to ask about studying MBA at Synetic.