ICP 4 - Joshmitha307/Python GitHub Wiki
Name : Joshmitha Tammareddy
Class id : 36
Mail : [email protected]
Aim :
1 . Implementing Naïve Bayes method using scikit-learn libraryUse iris dataset available in https://umkc.box.com/s/pm3cebmhxpnczi6h87k2lwwiwdvtxyk8Use cross validation to create training and testing partEvaluate the model on testing part
-
Implement linear SVMmethodusing scikit libraryUse the samedataset aboveWhich algorithm you got better accuracy? Can you justify why?
-
use the SVM with RBF kernel on the same dataset. How the result changed?
Code Explanation :
sklearn is a library which contains all the algorithms and datasets. Using sklearn we import all the algorithms and datasets and matrics, which is a marix containing all the elements in a dataset, and train_test_split from sklearn.model_selection to split the data and perform training and testing on it.
.png?raw=true)
The datasets are loaded into Iris and Naive Bayes model is fitted to the data. The dataset is considered as iris.data which is X and iris.target which is Y and prediction is made. The expected result is iris.target and the predicted is iris.target.
.png?raw=true)
Cross validation is done to compare the predicted and expected results. A matrix is used to find the accuracy of the classification. Both X and Y are split into training and testing data sets and a test size is given and also the random state is given so that we can understand it easily.
.png?raw=true)
The model is then then trained on testing and training data set and the model is evaluated on testing part.
.png?raw=true)
Linear SVC model is imported from the sklearn library. The data is divided into X and Y. The SVM model is fitted to the data.
.png?raw=true)
The model is trained on testing set and prediction is made based on X and Y.
.png?raw=true)
Cross validation is done and the model is evaluated based on testing part.
.png?raw=true)
Accuracy score is also imported from sklearn library.
.png?raw=true)
The datasets are loaded and split and are trained and tested.
.png?raw=true)
RBF model is fitted to the training sets. And prediction is made based on X_test and the accuracy score is printed.
.png?raw=true)
Output :
The Guassian model accuracy is 93.33%
The accuracy of SVm classifer on training set is 0.98 and on testing set is 0.97
The RBF Kernel accuracy score is 0.9
Conclusion :
By using SVM with the RBF Kernel accuracy has decreased.