Difference between revisions of "...predict categories"
| Line 8: | Line 8: | ||
If you ... | If you ... | ||
| − | * ...have two-class classification | + | * ...have two-class classification... |
| − | * | + | ** fast training, linear, then try the [[Perceptron (P)]] |
| − | * | + | ** fast training, linear, then try the two-class [[Naive Bayes]] point machine |
| − | * | + | ** linear, greater than 100 features, then try the [[Support Vector Machine (SVM)]] |
| − | + | ** fast training, accurate, and can have a large footprint, then try the [[Boosted Decision Tree]] | |
| − | * ...have multi-class classification | + | * ...have multi-class classification... |
| − | * | + | ** linear, then try the [[Naive Bayes]] |
| − | * | + | ** fast training, linear, then try the [[Logistic Regression]] |
| − | * | + | ** fast training, accurate, then try the [[Random Forest (or) Random Decision Forest]] |
| − | + | ** accurate, then try the [[Decision Jungle]] for multi-class classification | |
| − | * ... | + | ** ... accurate, can allow long training times... |
| − | * ... | + | *** ...then try the [[Deep Neural Network (DNN)]] |
| − | + | *** which is a type of is predecessors... [[Feed Forward Neural Network (FF or FFNN)]] and [[Artificial Neural Network (ANN)]] | |
Revision as of 19:35, 4 June 2018
Classifiers are ubiquitous in data science. The world around is full of classifiers. Classifiers help in identifying customers who may churn. Classifiers help in predicting whether it will rain or not. Classifiers help in preventing spam e-mails. If the targets are designed to be binary (two-class classification) then a binary classifier is used, the target will only take a 0 or 1 value.
___________________________________________________
If you ...
- ...have two-class classification...
- fast training, linear, then try the Perceptron (P)
- fast training, linear, then try the two-class Naive Bayes point machine
- linear, greater than 100 features, then try the Support Vector Machine (SVM)
- fast training, accurate, and can have a large footprint, then try the Boosted Decision Tree
- ...have multi-class classification...
- linear, then try the Naive Bayes
- fast training, linear, then try the Logistic Regression
- fast training, accurate, then try the Random Forest (or) Random Decision Forest
- accurate, then try the Decision Jungle for multi-class classification
- ... accurate, can allow long training times...
- ...then try the Deep Neural Network (DNN)
- which is a type of is predecessors... Feed Forward Neural Network (FF or FFNN) and Artificial Neural Network (ANN)
