Data-mining, Forecasting the future in the most accurate way

***This idea is scratched, Greg thinks it is silly***

Data-mining, Forecasting the future in the most accurate way possible in a world of apparent noise

Copyright 2015 Usmar Padow (amigojapan) usmpadow@gmail.com

-Take samples of all possible data over as long a period of time as possible(the more databases you have , the more data you will be able to mine from them) (an example of something you might want to data-mine is how much excursive you have done each day, if your goal is to loose weight, you may ask want to keep a database of what foods you eat, and everything you think may be related to weight gain/loss)


-Calculate Seasonal data. Usually the four seasons of the year(but not always)
(This is the step I least understand how to figure out)
(the most clearly defined “seasons” we know are the 4 seasons of the year, it would be easy to see this in the fact that our consumeristic society buys more goods during Christmas time, so if our goal is to determine how much of a product we will sell next year, then, looking at the sales pattern over the 4 seasons of the year and projecting it into the future seems reasonable to me)(but, there may be other patterns in nature(by nature I don’t only mean natural processes, man made processes are included) that follow cycles that are for example weekly or hourly and I think it should be possible to find a pattern in the “noise” of data we have, if a pattern really exists., there should be an algorithmic process we can use to find these patterns (although at the current time I still don’t know it)


-Determine your goals(your goals could be something like losing weight)

-Look for correlations between the seasonal data and your goals
(if for example, you see that eating less fat is correlated with loosing weight especially during the summer, you might have found a pattern)

-Remember that correlation does not always equal causation (but sometimes it does) (if you find that there is an apparent correlating between how many times you went to a homeopathic doctor and how much weight you lost, there may not be a real relationship between both things,   on the other hand,   perhaps the excursive you did in order to go to the doctor may turn out to be something that was related after all)

-Find a pattern
8I would like to be an expert and give everyone the algorithms necessary to do this,  unfortunately I am not very clear on how it is done myself,   but I know it is possible for things like finding patterns in automobile sales over a year) and I think it should be possible to find patterns in things that are not so clearly determined as automobile sales, like for example in the movements of the stock market, or in weight loss due to excursive and diet.

-Make a projection into the future

Remember that the further into the future you project, the less likely the forecast is likely to be accurate, since variables start changing.

Remember that a forecast is just a forecast, and may not be accurate