ud120-projects

This repository shows my solutions to the projects proposed in the course of Udacity's Intro to Machine Learning. The following are the answers for each section:

Naive Bayes

> python nb_author_id.py
no. of Chris training emails: 7936
no. of Sara training emails: 7884
Training time: 1.52 s
Predicting time: 0.209 s
The accuracy of Naive Bayes author identifier is 0.9732650739476678

SVM (Support Vector Machine)

• With training and test datasets at 100%, and using a linear kernel:

> python svm_author_id.py
no. of Chris training emails: 7936
no. of Sara training emails: 7884
Training time: 111.625 s
Predicting time: 10.488 s
The accuracy of SVM author identifier is 0.9840728100113766

• With training and test datasets at 1%, and using a linear kernel:

> python svm_author_id.py
no. of Chris training emails: 7936
no. of Sara training emails: 7884
Training time: 0.05 s
Predicting time: 0.376 s
The accuracy of SVM author identifier is 0.8845278725824801

• With training and test datasets at 1%, and using a rbf kernel:

> python svm_author_id.py
no. of Chris training emails: 7936
no. of Sara training emails: 7884
Training time: 0.049 s
Predicting time: 0.708 s
The accuracy of SVM author identifier is 0.8953356086461889

• With training and test datasets at 1%, using a rbf kernel and C = 10000.0:

>python svm_author_id.py
no. of Chris training emails: 7936
no. of Sara training emails: 7884
Training time: 0.048 s
Predicting time: 0.804 s
The accuracy of SVM author identifier is 0.8998862343572241

• With training and test datasets at 100%, using a rbf kernel and C = 10000.0:

>python svm_author_id.py
no. of Chris training emails: 7936
no. of Sara training emails: 7884
Training time: 122.881 s
Predicting time: 16.143 s
The accuracy of SVM author identifier is 0.9960182025028441

• With training and test datasets at 1%, using a rbf kernel and C = 10000.0. What are the predictions for 10th, 26th and 50th element?

> python svm_author_id.py
no. of Chris training emails: 7936
no. of Sara training emails: 7884
Training time: 0.041 s
Predicting time: 0.833 s
The accuracy of SVM author identifier is 0.8998862343572241
Prediction for element 10: 1 #Chris
Prediction for element 26: 0 #Sara
Prediction for element 50: 1 #Chris