Scribd
Upload
21 views2 pages

AIE Portfolio3

Uploaded by

huyquangph2004
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
21 views2 pages

AIE Portfolio3

Uploaded by

huyquangph2004
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Nam: Phan Huy Quang

Student ID: SWH00986 or 104177128


Studio class attend: Tuesday afternoon, 3-5pm

Summary Table of Studio 3: Activity 6

SVM model Train-test Cross validation Cross


split validation
mean

Original features 89.17% 90.28%, 89.42%, 89.25%, 88.39%, 89.6%, 89.18%


89.6%, 88.05%, 89.34%, 90.03%, 87.87%

With hyper parameter 84.12% 83.83%, 84.61%, 84.09%, 84.26%, 84.52%, 84.26%
tuning 84.61%, 84.26%, 84.18%, 84.44%, 83.82%

With feature selection and 85.61% 85.12%, 86.50%, 85.30%, 85.30%, 85.73%, 85.59%
hype parameter tuning 86.07%, 84.95%, 85.64%, 85.98%, 85.28%

With PCA and hyper 84.20% 84.36%


parameter tuning 84.01%, 84.69%, 84.09%, 84.26%, 84.61%,
84.61%, 84.26%, 84.35%, 84.69%, 83.99%

Summary Table of Studio 3: Activity 7

Model Train-test Cross validation Cross


split validation
mean

SVM 89.17% 90.28%, 89.42%, 89.25%, 88.39%, 89.6%, 89.18%


89.6%, 88.05%, 89.34%, 90.03%, 87.87%

SGD 69.91% 85.98%, 85.12%, 88.91%, 88.48%, 87.41%


82.55%, 88.99%, 87.88%, 88.56%,
89.68%, 87.95%

Random Forest 90.51% 93.90%, 92.43%, 92.86%, 92.09%, 92.59%


93.04%, 93.38%, 91.75%, 92.09%,
93.55%, 90.79%

MLP 77.53% 85.81%, 83.15%, 84.18%, 87.70%, 84.63%


87.19%, 88.74%, 87.88%, 71.88%,
80.31%, 89.50%
Source code:
My source code can be accessed through: AIEportfolio3.ipynb
When you run this code, files will be created:
● Step 1: Data collection:
○ Filename: ‘combined_data.csv’
○ Path: ‘/content/combined_data.csv’
● Step 2: Create composite columns
○ Filename: 'composite_data.csv'
○ Path: ‘/content/composite_data.csv’
● Step 3: Data pre-processing
○ Filename: ‘Preprocessed_data.csv’
○ Path: ‘/content/Preprocessed_data.csv’

Step 4: Training outcome:

Model Train-test Cross validation Cross


split validation
mean

SVM with original features 75.07% 79.34%, 76.67%, 75.00%, 77.50%, 70.83%, 77.10%
77.50%, 80.00%, 77.50%, 75.00%, 81.67%

SVM with hyper parameter 75.07% 75.21%, 75.00%, 75.00%, 75.00%, 75.00%, 75.19%
tuning 75.00%, 75.00%, 75.00%, 75.83%, 75.83%

SVM with hyper parameter 75.62% 77.69%, 80.00%, 82.50%, 72.50%, 66.67%, 75.69%
tuning and 10 best features 65.00%, 75.00%, 73.33%, 87.50%, 76.67%

SVM with hyper parameter 75.07% 75.21%, 75.00%, 75.00%, 75.00%, 75.00%, 75.19%
tuning and 10 principal 75.00%, 75.00%, 75.00%, 75.83%, 75.83%
component

SGD with original features 81.44% 75.21%, 72.50%, 71.67%, 84.17%, 70.00%, 74.35%
72.50%, 80.83%, 66.67%, 73.33%, 76.67%

Random Forest with 89.47% 85.12%, 92.50%, 90.83%, 89.17%, 88.93%


original features 78.33%, 90.83%, 95.00%, 91.67%,
91.67%, 83.33%

MLP with original features 85.32% 89.26%, 83.33%, 84.17%, 72.50%, 81.76%
78.33%, 89.17%, 85.83%, 78.33%,
76.67%, 80.00%

Step 5: Model selection:


1) SVM model with hyper parameter tuning and 10 best features is the best SVM
model as its score is higher than others
2) Random Fores is the best model as its score is higher than others

You might also like