How true are predictions based on Big Data?
It is a very difficult question to solve because it is necessary to take into account when working making estimates of what can happen based on Big Data. Everything that is obtained from Big Data is based on past data that has already happened, but the future is not determined, especially when it comes to people’s reactions.
People are analogical, not digital, and in many cases hardly predictable, because the variations in the conditions of our environment make us react differently than we did before.
Someone will argue that that’s what Artificial Intelligence is for, but I think AI doesn’t solve everything.
On the one hand there are the large numbers of statistics that are generally fulfilled with greater or lesser degree of proximity. From them we can estimate the accidents that will occur, the cases of cancer that will appear, and a multitude of other general events; but one thing is to generalize and another very different to particularize for certain specific people.
The case in France in 1965 is known worldwide. The lawyer André-François Raffray agreed to pay a 90-year-old woman a monthly contribution of 2,500 francs until the day of her death, to inherit at that time the flat she used. It seemed to be a good business because then the life expectancy of a woman in France was 74.5 years.
The woman (Jeanne Calment) lived 32 years more and Mr. Raffray died before her, paying for the floor twice what it cost in the market.
Conclusion: what can be a good business applied to many people ( that’s what banks do ), can be a disaster for a particular case.
Specializing is very difficult, and Big Data is useful for general cases and behaviors, although knowing that these behaviors can change significantly due to abrupt changes in the environment.
Therefore, the Big Data forecasts according to what things apply can be more or less accurate and more or less relative.
Individual people are unique and unrepeatable, and I am glad that it may remain that way for many years.
.
