por Black box models are those that describe the functional relationships between the inputs and outputs of a system..
They are called black box models because it is difficult to access their inner workings, that is, their structure, logic and internal components are impossible to elucidate.
Fields of application of black box models
There are different areas in which black box models can be applied, including science and technology, but above all these models are (or were, we will see their drawbacks later) widely used in the financial market to execute trading operations machine learning related to predictive analytics.
These models contained algorithms that could change the course of the decision-making process in a company with some investment, since the future actions to be taken depended on their results.
Added to this, black box models also They are used as models of consumer behavior theory to observe and discover consumer behavior and stimulus/response patterns. That is, in the area of marketing and analysis (not prediction) of the market it is very useful, even today.
Disadvantages of black box models
Due to their character, black box models have lost popularity over the years, since more than a help They become gibberish that ends up being an inconvenience for investors in the long term.
Strategies involving black box models are quite problematic and their use depends largely on market conditions and the market cycle. When the stock market begins to have imbalances, becomes destabilized and becomes an element of extreme volatility, it is not advisable to use this type of models because they can cause a lot of losses and dangers in this area.
However, despite their drawbacks, black box models have once again acquired a hybrid character in which they have both detractors and admirers. This is, in a sense, beneficial, since current advances by disciplines such as data science, machine learningg and other related fields have made this type of models a more than viable alternative when requiring predictive analysis.
Peng Zhang in chapter 19 of his book Advanced Industrial Control Technologytitled Industrial control system simulation routines, states that Black box models are very useful if what you want is simply to faithfully represent some trends in the behavior of some process.
It also mentions that there are two classifications for black box models: linear models and nonlinear models.
Linear and nonlinear models
Linear models are those transfer function and time series models. One of the most used techniques to identify the parameters of black box models is least squares.
Nonlinear models, for their part, are those in which neural networks predominate. The use of these in the construction of models has made possible the high economic availability that exists today in the market for some computing accessories.
White box models
There are, contrary to black box models, white box models. These models have all their internal mechanisms free of analysis and can be evaluated and observed at any time required. They are sometimes called transparent box or glass box models, due to their clarity of processes and responses.
Whats Next?
Now that you have learned how black box models work and what white box models are, let’s continue learning together! At Keepcoding we have a great learning offer for you: through our Big Data, Artificial Intelligence & Machine Learning Full Stack Bootcamp you will be able to train theoretically and practically in everything you need to become an expert in the IT sector. Go ahead and request more information now!