Is it possible for anything to statistically analysed?

Today I want to discuss the possibility of statistically analysing every issue in the real world. Firstly, it is possible to analyse every issue in the real world; however, the real problem is how accurate and useful the results will be. I also believe in the future, when more information is available, technology is developing, it will be more accurate to conduct statistical data analysis.

I do have a rough method for computers to conduct data analysis for interested matters without too much manual interaction; however, I am not a computer programmer, so I do not have the professional skills or knowledge to discuss very specific problems in my method. I do believe in the future, more information will become available on the Internet and if computers are fast enough at the time, they will be able to collect all the information from the Internet. Computers need to find out the correlations between different factors, as well as the significances of the correlations. Because there are countless factors, it is impossible for computers to find all relavent variables according to the input interested variables and a continuous practice and a storage of correlations between variables are required. Computers should be like AlphoGo to practice a lot and improve its own abilities to find out the most accurate correlations between variables. Once an interested variable is inputted, the computer can use its storage to find out relevant variables and use the relevant variables to form an equation to provide a statistical description of the interested variable.

Such process has several advantages. Computers can form more complicated transformation to fit the reality better than we manually fit the models. In addition, once computers spot relevant variables and store the information and will never omit these variables; however, when we manually design models, it is very possible for us to omit some important variables. More importantly, AlphaGo has shown to us that AI can improve itself, so I believe it is possible for computers to develop better-fitted regression models without our intervention.