“Garbage in and garbage out” or GIGO is a profound concept for any type of statistical analysis, either hypothesis testing or forecasting. Errors in data are profoundly frustrating when considering statistical decision-making or precision, and saying “we are going to need a bigger boat” does not solve the problem.
The scientific method involves the posing of a conjecture, the collection of information, and the examination of whether the information is consistent with the conjecture. If not, the process generally eliminates the conjecture. It is well understood, however, that one can wrongly eliminate a conjecture that is, in fact, true. As a result, a key to this problem is the collection of information. If something is mis-measured, this logical process fails. Put simply, it is better to observe a small share of the universe precisely than to observe the whole universe imprecisely.
Real estate data may, for example, be contaminated with measurement errors. Why, you ask? What is measured and how it is measured may change over time depending on the interests of users. For example, in commercial real estate, it may be not possible to have a consistent rent or vacancy series in office space. One might try to average or impute rents, but one doesn’t know tenant improvement expenses. As a result, one cannot calculate basic Net Operating Income (or economic rents).
At Forty5Park, we understand these issues, and the dangers associated with them, very clearly. Both our data science and data engineering teams, assisted by me, are intently focused on these issues. Combined with our probabilistic approach to forecasting, we are well-placed to deliver on data-driven market intelligence.
So does GIGO still rule? It never stopped.
