One of the first words we associate with the word future results uncertainty.
In the past we relied on vicars of deities, oracles, drugs, or a mixture, to get some answers about the future.
Today, finding ourselves in the scientific age, closing the Comtian reference, we have various and different ways of measuring uncertainty, in a methodical, verifiable, repeatable way.
In analytical chemistry we have measurement uncertainty, of an instrumental nature. In statistics we have a more particular uncertainty, called variance, which originates in the calculation of probabilities and manifests itself as probability distributions. For example, the bell curve or normal curve is very well known, as various natural phenomena have this distribution, for example the height of people, their IQ, instrumental error, etc.
An immediate explanation of variance: average fluctuation from the average, but let’s try to see more concrete explanations.
Think about the work of building a handcrafted object, such as a chair. Here the variance becomes the precision in the creation of the parts. If all the chair legs are exactly the same length, we will have low variance. But if the leg lengths vary greatly, the variance will become high.
Imagine you manage an office and take care of employee arrival times. The variance could represent the variability in the times in which employees arrive at work. If everyone always arrives at exactly the same time, the variance will be low. But if you observe large differences in arrival times, the variance will be high, indicating greater uncertainty and unpredictability.
Part of statistics studies variance. Constant or homogeneous phenomena are not of interest to statisticians or data scientists. Statistical models, in fact, explain the variance of an objective variable, not exactly, but always bringing an error. Usually normally distributed.
The models seen at school, for example in physics lessons, have no errors. The error term is found in experimental physics, because there we have instrumental errors. The equation or model of the area of the square does not have a term indicating the error. The error can emerge when we take the measurements of the sides, therefore when we want to apply the formula after using a measurement tool.
Statistics therefore allows us, through modern calculators, to understand which variables influence the uncertainty in the objective variable. For example, through statistical tests, which give statistical significance. But not without problems. It is not just a question of the error in the model but of issues already known thanks to analytical philosophy, which, unlike other schools of philosophical thought, has practical implications. Examples: the inductive turkey problem, Simpson’s problem, the nature of causality (a term I use with caution), measurement theory.
Let’s discover together which variables have influenced the past of your company, if conditions allow it, to understand how to best equip yourself for the present and the future, through a path that starts with a call. Or we can monitor the length of the legs over time, to reduce uncertainty, waste and problems.