use of the classification when solving a particular problem
It has been concluded from the all of the above discussion that the use of the classification is not that simple in terms of solving a particular problem. It has been identified that the use of classification approaches depends upon the type of the problems which the things are given for the classification in the groups of the items. The major purpose of the analyzing these results is the identification of the approach that has been the used as the most appropriate for the sake of solving a particular problem. Now, to identify the most suitable approach we have different techniques such as the use of evaluation metrics which identify the approach that have the most suitable accuracy and the most suitable results in terms of solving the defined problem. As we have already defined that the accuracy formula is available for the sake of identifying the evaluation metrics. This approach is commonly used for the sake of identifying that either the proposed model is better than the previously given approaches. The accuracy can be identified through different approaches. Now, to identify the most suitable formula it is important to understand the nature of the problem. It has been identified from the previously computed results of Machine Learning algorithms that the most suitable training accuracy has been obtained from Logistic regression. As we know that training accuracy is obtained from training set which is use for the sake of fitting the available model over the given scenario. In this data, there is proper labeling and it have appropriate identity. Moreover, this data can be defined as the seen that in which there is proper classification of the items. In addition, the most suitable f1 score has been obtained from Logistic Regression model. Likewise, the most suitable precision and recall values have been obtained from Logistic Regression as well. Hence, in conclusion it can be said that the most suitable approach for the sake of recursive features elimination is Logistic Regression Model. Hence, in sum it can be said that the most suitable approach for the sake of recursively feature elimination is the Logistic regression. Although the recursive feature elimination has its own working approach and it always use the approach of excluding the features through the model. Hence, the overall results obtained from the logistic regression can be said as the suitable in terms of having better accuracy of the model. Moreover, to identify the purpose that why the logistic regression have better accuracy in all of the evaluation metrics of the recursive feature elimination, it has been noted that when the model of the logistic regression is trained in terms of the finding the most suitable features and getting the suitable results, the overall cost function involved in the accuracy computation of the model is very low. Likewise, the model always predicts the most suitable results for the classification purpose. Moreover, the results obtained from this approach always contain the attributes that are most logical and important for the defined problem.
Moreover, in case of the Random forest, the approach has been used in all of the three techniques containing Support vector machine (SVM), Logistic Regression, kNN, and Naïve Bayes approach. It has been concluded from the results and the computation of the model that the most suitable accuracy in case of feature selection through random forest has been obtained from Logistic regression. The most suitable training accuracy has been obtained from the SVM classifier. The most feasible f1 score for the sake of performing the experiment is obtained from Logistic Regression. In addition, the better results in terms of precision and recall is obtained from Logistic regression. Hence, in sum, it can be said that the most suitable approach for feature selection using random forest technique is Logistic Regression model. In sum, it can be concluded that the most appropriate approach in both of the instances where random forest and recursive feature elimination has been used, this logistic regression based approach is there for the sake of computation of the desired results through the most suitable accuracy of the model along with the better accuracy of the training model.