HW3
98 views
Skip to first unread message
Roman Hess
Mar 21, 2023, 3:50:44 AM3/21/23
to NCCU DS4CS
Dear Professor,
for this week's assignment, I would kindly like to ask aquestion.
What kinds of Feature Engineering do you expect to see in this homework? Could you give an example?
Thank you very much for your support.
Kind regards
Roman Hess
Mike Hsiao
Mar 21, 2023, 10:00:46 PM3/21/23
to NCCU DS4CS
Hi, Roman,
You may take a look at Orange's website, and there is some useful information.
For example,
There is a tutorial named "Feature Ranking"
2) Feature Statistics widget
3) Rank
4) Orange 10: Feature Scoring and Ranking
And do not forget that you may apply any 'preprocess'.
Even a little feature engineering might help. We wish you could try.
Please note that we have modified the homework as follows.
The due date has been extended as well.
----
The due date of H-NID has been extended to 3/27 (Monday).
We notice that the testing data provided by Kaggle (Test_data.csv 2.42 MB) has no 'target' column.
https://www.kaggle.com/code/sampadab17/network-intrusion-detection-using-python/input
You may use one of the ways to analyze such data in your homework.
1. Use k-fold analysis to obtain the accuracy (or F1 score) value.
2. Split the training data into two parts. Use one of them for training and use another for testing.
(Hint: there is a random sampling widget in Orange)
3. Use the training data for training, and use testing data for prediction (and there has no correct answer for your referencing)
4. Apply unsupervised learning algorithms (k-means, clustering) on both training and testing data.
Any of the above is okay. The purpose of this homework is to help you to get familiar with data analysis flow.
We notice that the testing data provided by Kaggle (Test_data.csv 2.42 MB) has no 'target' column.
https://www.kaggle.com/code/sampadab17/network-intrusion-detection-using-python/input
You may use one of the ways to analyze such data in your homework.
1. Use k-fold analysis to obtain the accuracy (or F1 score) value.
2. Split the training data into two parts. Use one of them for training and use another for testing.
(Hint: there is a random sampling widget in Orange)
3. Use the training data for training, and use testing data for prediction (and there has no correct answer for your referencing)
4. Apply unsupervised learning algorithms (k-means, clustering) on both training and testing data.
Any of the above is okay. The purpose of this homework is to help you to get familiar with data analysis flow.
----
Thanks,
HSIAO
Roman Hess 在 2023年3月21日 星期二下午3:50:44 [UTC+8] 的信中寫道：
Reply all
Reply to author
Forward
0 new messages