these days the are so many algorithms to predict future events such as failure of machines or electrical components degradation.
which method is your favorite and why?
can you also provide a source code if there are any available online?
Dear Sajad Saraygord Afshari
Linear regression may be one of the most famous and well-understood algorithms in statistics and machine learning. Predictive modeling is mainly related to minimizing error model or making the most accurate predictions, at the expense of explanation.
For more, Kindly look: https://builtin.com/data-science/tour-top-10-algorithms-machine-learning-newbies
Kind Regards
Qamar Ul Islam
Dear Sajad Saraygord Afshari,
Several Machine/Deep Learning algorithms are available in the literature. The selection of the best ML algorithm for a particular problem depends on many factors such as the availability of data for training, types of system/environment, availability of hardware for processing, etc. However, for prediction and decision-making problems Reinforcement learning algorithms are best (according to my observations).
For more detail, you can check
Article Machine learning-based physical layer security: techniques, ...
You can check the links for source code in Matlab
1) https://in.mathworks.com/help/reinforcement-learning/matlab-environments.html
In Matlab Simulink Deep reinforcement learning toolbox is inbuilt. You can check the following link
https://in.mathworks.com/products/reinforcement-learning.html
There is no a single tool which is the best in very case. There are a lot of articles proving the overperformace of a given tool (chosen or improved or invented one) above other tools but... applied for solving a certain problem. Which one will be the best depends on:
- the problem itself
- number of independent variables
- types of variables (e.g. some tools do not accept categorical variables)
- number of cases we can use for training process
- the nature of the problem (sometimes it is highly uncertain e.g. dependent on the human choice, which is not rational in every case)
- etc.
So there is no a single, the best tool.
Hi, Sajad Saraygord Afshari
There is no direct answer to conclude which ML/DL algorithm is the best for prediction. It depends on the application and specific requirements. For instance, in some applications it is required a fast training and testing speed; however, in some other applications it is required a high prediction accuracy (or a low computational cost). There is always a trade-off when we consider the ML/DL algorithms.
Regards,
Zhenyu
There cannot be any favorite algorithm/model, as it depends upon numerous factors, the most important of which (as per me at least) is 'what inductive bias are you willing to bargain', which in turn depends upon the dataset. Almost every model leverages distinct assumptions upon the data, thus performs variably, in accordance to the scenario.
Other factors have already been discussed by others in this discussion.
- Badges
- Science topic
- Similar topics
- Automotive Engineering
- Machines