Comparing Actual to Estimated Sales

We offer several tools that facilitate comparison between estimates generated by predictive models and actual sales figures:

comparison actual versus estimated sales

comparison actual versus estimated sales

Numeric tables detailing the daily model error, the 5-day cumulative model error, and the running-total model error are always available online. Simply click on the "Include Error Table" checkbox at the top of the screen when you view your saved models.

Online tools provide access to scatterplots, bar charts, and other graphing tools that help visualize any discrepancies between actual and estimated sales.

Access historic, stored predictive models using the online regression tool. If you haven't already used the tool to analyze datasets, here is a screenshot.

In certain situations, such as in the example where we assessed the effectiveness of a marketing effort, we use offline tools to overlay predicted versus actual sales, with explanatory text added. For those interested in raw numbers, we provide the sum-of-squared errors and the proportion of variability in sales we are able to account for with the model. We are also happy to disclose results of significance tests in order to support the inclusion of predictor variables. We use conservative post-hoc analysis (simple Bonferroni correction) to ensure that all variables in our cross-validated model is truly significant at the p < 0.05 level.

Use the online regression tool to examine our models, or build your own...

Statistical Analysis and Modeling Demo

Modeling techniques include nonlinear multiple regression, binary or multinomial logistic regression, and canonical analysis.

Exploratory statistical modeling is used to discover which variables are associated with sales

Free business advertising	Home \| Privacy Policy \| Contact Us
	Copyright © 2006 BrainTech, LLC. All Rights Reserved