i have some basic knowledge in AI and machine learning, but a bit confused solving a concrete problem.
i have the following scenario: Given are features and labeled data (0 or 1). I want to predict the probability, new data takes 0 based on the feature values of this new data.
I know this is supervised learning, but what method can I use for predicting here (i think logistic regression or neural networks should be an option?) and if theres any preimplemented libaries in Python I can just fed the data to and it will presict?
CodePudding user response:
For both quick overview and example implementations, please visit scikit-learn estimators. Your task is more likely of classification, which you can think it as how it's possible to fit (from 0.0 to 1.0) to a specific category.
There are vastly available models to use. Most of them work through minimizing (or iterative optimizing) the cost function to obtain a valid prediction model. In particular cases, E.g., low dimensional feature, a decision tree which is one of the primitive algorithms, also can give reasonable results.
I would recommend that you dig further into how well the model can be generalized
and interpretable
; both are just as important as model accuracy.
CodePudding user response:
Given the nature of your problem, there are many supervised learning models that you can use. Here are some of the algorithms with what you want to achieve.
Do you prefer Speed or Accuracy? For Accuracy Kernel SVM Random Forest Classifier Neural Networks Gradient Boosting Classifier
For Speed, is your data explanable? If yes, use Decision Trees Logistic Regression
If not, use Naive Bayes if dataset too large Linear SVM if dataset is small
These are for Supervised learning (Classification), for other forms of learning, there are other models as well. You can view this link for more information.