Pricing and Revenue Management


Pricing and Revenue Management

This exercise is based on a sample provided by a fuel pricing solution provider. Imagine how a pricing analyst and helping a US petrol retailer to analyze its pricing strategies. A sample of data was collected from one of sites this retailer operated. You shall produce a business analytical report to help the client understand its demand structure, modelling the pricing decisions and provide insights into the pricing strategies.
The data in the file shows Petrol sales in gallons and prices in dollars. The columns are as follows:
• Date
• Weekday
• Vol – the volume in gallons of regular unleaded petrol sold at the site on that day
• Price – the price per gallon in dollars changed by that site
• AvgCompPrice – The average price charged by competitors to the site
• MinCompPrice – The lowest priced competitor to the site on the given day
• MaxCompPrice – The highest priced competitor to the site on the given day
Marking principle:
1. Understanding on the retail fuel (petrol) market and provide prescriptive analytic insight into the pricing decisions in such as a business context. (10%)
Note: You may want to associate spot market price with the retail fuel pricing. Some market information and data can be found in websites like ( You should be able to use summary statistics to provide descriptive analytics and insights for the pricing data, and ask relevant research questions and hypotheses which will be studied in the following analysis.

2. Fit the price-response demand model and compare different demand models, analyze the own price elasticity and competitor’s price elasticity, and more importantly provide insights into petrol pricing using the predictive analytics. (45%)
Notes: You should compare at least the following demand models and discuss the corresponding hypothesis behind the model.
a. Linear regression with volume (sales) as the response (dependent variable) and own price and average competitor price as the predictors (independent variables).
b. Linear regression with natural logarithm (LN) of volume as the response and LN(price) and LN(average comp price) as the independent variables.
c. Include the weekday as a predictor by creating dummy variables (Note that you need to remove linear dependency).
d. You can also include the month or season as the predictor to address the seasonality effect.
e. You can further improve the model performance by examining the reference price effect and competition effect. You can then measure the competitive price elasticity.
f. Compare the above models to other model specifications you are interested.
Write down the model coefficients and R squared output (and adjusted-R squared output), then compare the goodness-of-fit using the (adjusted-)R squared output. Next, choose the best performance model and write down the demand function. You should also provide the model diagnosis to ensure that the model is specified properly. Finally, comment on the price elasticity of the sales and the impact of different characteristics (e.g., weekday, seasonality) on the price elasticity. You should be able to interpret the results clearly. As a business report, only presenting the statistical results is not enough. More importantly, you should use the statistical results (including the comparison of the models) to provide insights into the pricing strategies (even before you formally formulate the optimization models).

3. Propose the pricing optimization models and derive the optimal pricing strategies.  be able to state the model formulation clearly, explaining the meanings and implications of each term. Then  show how  to develop the optimal solutions and analyze the characteristics of the pricing solutions. Finally,  be able to examine the performance of the proposed pricing solutions using the sample data. Since the cost information is not available. use the regular gasoline spot price in the same period of time as the proxy ( For those days without spot price information (as there was no trading in spot market), using the available price of the previous day. (40%)