Let’s continue to look into the problem of using averages by introducing Anscombe’s Quartet. In 1973, statistician Francis Anscombe assembled this now famous quadrant that proves to that you can have data sets that have radically different graphs, yet have the same exact summary statistics. The four graphs below have the same number of data points, averages, variance, correlation and trend line. Yet the graphs look completely different. There’s not much that I can say about the implication of this for hotel analysts that you can’t figure out for yourself, except to note that Anscombe elegantly proves that it is negligent to use measures like PAR, ADR, Caputre, etc., without exposing the underlying patterns in these numbers (a central motivation behind this blog). The morale here is to not only graph your data to make sure there’s no hidden, counter-intuitive trend, but also to always look at the multi-dimensional patterns that affect your summary statistics. To see Anscombe’s original data set go here.
Don't Miss the Next Revenue Discovery Insight. Subscribe Now
Robert Hernandez, Statistical Analysis and Data Mining for Revenue Growth Robert is an expert in the field of mathematical Hotel Optimization and Analytics. He has spent the last 17 years building data-driven forecasting and optimization models for companies in over 20 different industries, from tech to tourism. Robert possesses a very unique skill set including cross-disciplinary experience, advanced mathematical and analytics skills, data transformation, industry-specific knowledge and business-process improvement expertise. Robert began his career at the Walt Disney Company in Revenue Planning. Read More+