Regression Analysis for Business Insight - Consulting Case

Explore ordinary least squares estimation and Gauss-Markov assumptions, deriving the best linear unbiased estimators from a simple linear regression with one explanatory variable, and assessing residuals.

Apply ordinary least squares to identify sales drivers, estimate marketing roi and price effects, assess competitor impact, seasonality, and future scenarios for 10% efficiency gains and 20% marketing spend changes.

Explore how to structure weekly sales data for regression analysis, including date identifiers, seasonality dummies, client and competitor marketing, promotions, pricing, and macro indicators.

Assess how the data dictionary's impact column frames causality versus correlation, detailing monthly seasonality, marketing variables, pricing, promo effects, and macro conditions.

Identify seasonality in sales data across multiple years, using pivot tables and correlation analysis to ensure accurate coefficients and informed business decisions.

Explore a time series regression model for business insights, detailing how sales at time t decompose into a constant, twelve explanatory variables with corresponding beta coefficients, and an error term.

Examine the model output from a python-based analysis, verify coefficients, standard errors, t statistics, p-values, and confidence intervals, and assess causal impact versus correlation.

Analyze regression model statistics such as R-squared, Durbin-Watson, and MAPE, then perform a model decomp to quantify daily incremental effects of marketing, price, and competitive variables on sales.

Analyze the actual versus predicted graph (AVP) to compare red actual sales with blue predicted data and inspect residuals from the decomp tab for daily model performance.

Explore how to analyze residuals to validate regression models, check for autocorrelation with Durbin-Watson, assess homoscedasticity, and test normality via histograms and Q-Q plots to ensure Gauss-Markov conditions.

Explore residual analysis to validate regression models, checking residuals for randomness, independence, and constant variance, and using scatter plots and normality tests to assess Gauss-Markov and OLS estimator properties.

Use regression insights from the DICOM and decomp tabs to answer client questions on 2022 marketing impact (112,000 units, 22%), competitor marketing impact (-10,272 units), and August 2019 baseline sales.

Analyze the 2022 marketing ROI, rising to about 1.74 from 2021, using unit sales and a $60 price, and note the competitor’s ROI losses of -0.94 in 2022.

Apply regression analysis to pricing questions, estimating units lost and revenue under price changes, and assess elasticity of demand with a 10% price increase scenario.