"Abstract (0.75/0.75)
Probably try to make it shorter. You don’t need to go in detail of what you are going to do. -Yicheng
Background (1/1)
Problem Statement (1/1)
Data (0.75/1.25)
You need to present your cleaned data and demonstrate that your data is usable. You might also want to include EDA so that you can justify some of your cleaning process (e.g. why do you keep only top 20 of 1786 unique device software?) Also, the merging process seems to be invalid ---- they have different number of observations, thus you cannot directly merge them.
-Yicheng
Proposed Solution (1.25/1.25)
Metrics (1.25/1.25)
Preliminary results (1.25/1.5)
Benchmark model should be pre-existing models (models created by someone else). Or, if you really cannot find one, you can compare and contract your models (don’t go one vs. rest, go one vs. one) (e.g. SVM vs. KNN vs. decision tree instead of SVM vs. KNN, SVM vs. decision tree) -Yicheng
Rememer to show your work here. -Jason
Ethics & Privacy (0.5/0.5)
Team expectations(0.25/0.25)
Timeline (0.25/0.25)
Proposal grade change:
8.25->8.25
Other comments:
You might want to perform EDA on you dataset to have a deeper understanding of your data.
"
Total: 8.25/9
"Abstract (0.75/0.75) Probably try to make it shorter. You don’t need to go in detail of what you are going to do. -Yicheng
Background (1/1)
Problem Statement (1/1)
Data (0.75/1.25) You need to present your cleaned data and demonstrate that your data is usable. You might also want to include EDA so that you can justify some of your cleaning process (e.g. why do you keep only top 20 of 1786 unique device software?) Also, the merging process seems to be invalid ---- they have different number of observations, thus you cannot directly merge them. -Yicheng
Proposed Solution (1.25/1.25)
Metrics (1.25/1.25)
Preliminary results (1.25/1.5) Benchmark model should be pre-existing models (models created by someone else). Or, if you really cannot find one, you can compare and contract your models (don’t go one vs. rest, go one vs. one) (e.g. SVM vs. KNN vs. decision tree instead of SVM vs. KNN, SVM vs. decision tree) -Yicheng Rememer to show your work here. -Jason
Ethics & Privacy (0.5/0.5)
Team expectations(0.25/0.25)
Timeline (0.25/0.25)
Proposal grade change: 8.25->8.25
Other comments: You might want to perform EDA on you dataset to have a deeper understanding of your data. "