In statistical analysis and data mining projects, building a demand curve is about finding a good approximation of the relation between price and quantity. The relation between the price and quantity variables can be derived from sales and income books and usually retrieved form the organizational computers or data base. At this point all you know is what the sales levels for a price list are for each one of your products and this article we would like to give you the tools to provide a sales figure for different prices of the same product (prices that were not assigned to your product before).

As an example (below) are some product price and quantity figures for a year including special campaigns and different sales channels.

Viewing the table it is easy to find out that there is a decreasing of sales quantities when price is rising. However, can you answer the question: what will the sales quantity if the price is $83?

To get a better answer to this question you can build a demand curve that will help you to get as precise figures as possible. Building a demand curve using Analysis Studio software is easy and can be done by executives and can be a good first step for moving from just viewing data to data mining and predictive analysis.

Step No .1: Selecting the demand curve variables

The Explained variable (in this case the Quantity) is the variable that we would like to know how changes in the explanatory variable values (In this case: price levels) affect it. As "Explanatory variable", select "Price" variable and move it from "Explanatory Variable" frame to the "Selected Column" frame.

Analysis Studio is capable of finding the best-fit regression for the given problem by using the "All Regressions" option. Once all regressions are selected, 5 different regression models will be computed and the best (fit) one will be automatically selected. You may change the Analysis Studio selection and select a different regression model. Click the "Next" button.

Step No .2: Viewing results

Above you can see the demand curve computed by Analysis Studio. The Analysis Studio regression wizard selected Polynomial regression as the most appropriate however you may choose other regression types from the computed regressions in the Manual Defining tab. Note that for some data sets not all regressions can be computed due to data quality and behavior. In our case, there is a big deference between polynomial regression and linear regression (as can be seen in the explanation value column in the image above). The Polynomial regression is relatively hard to interpret and you might want to select another regression model to use as the demand curve.

The Explanation value is the regression model ability to explain the variation of the explained variable (In our example the "Price" variable can explain 89.61% of the Quantity values). It seems like Price can be a very good predictive measure for "Quantity" in this data set.

Click the "Finish" button to view the model in the Analysis Studio main screen.

Note! Statistical knowledge is required in order to determine the regression model quality and a statistician or a qualified analyst should approve statistical and data mining models.

Step No .3: Working with model results

After computing an appropriate regression model, we can finally answer the question: what will the sales quantity if the price is $83?

The regression model shows a predicted value of 2392 units to be sold at price of $83.

Click the "What If" tab in the regression tab at the main panel of the Analysis Studio application. A "What if" scenario is a way to see the effect different variable values have on the explained variable. This is an important step in data mining and data analysis since it gives the analyst an independent aspect on the results and their limitations.

Step No .3: Creating a demand curve

After the regression model has been generated and verified an actual demand curve is built using the Analysis Studio "Sensitivity Table". Click the Sensitivity Table tab in the regression model tab.

The Sensitivity Table generates a set (series) of regression model results based on given values to predict demand and to understand the affects of different prices on sales levels. As can be seen in the image above, select the Explanatory variable and enter the minimal/maximal values to be computed. The increment field is the resolution of the curve between the minimum and maximum values.

The curve will have 1 + (Maximum – Minimum) / Increment points.

Next steps:

• For further information please visit the support section of our site.
• Download Analysis Studio free edition – Includes unlimited variables, up to 500 rows of data.
• Purchase Analysis Studio