jcasoli
commented
2 years ago Thanks for your comments @andytai7! In regards to the project proposal, I wasn't sure whether to make changes in the proposal or just ensure that we address them in the final report. I decided to go with the later option.
Here are some specifics:
What packages will you use for EDA what types of methods?
- We made mention in our final report that altair was used for EDA
What about class balance, there could be an imbalance in the classes, in which you would have to under sample or oversample. Which one will you utilize?
- We mention in the final report (in the EDA section) that we do have class imbalance, and therefore we decided to optimize the
class_weighthyperparameter of LogisticRegression. It turns out that setting class_weight="balanced" does improve model performance.
What about missing data, how will you handle the missing data?
- We used a
SimpleImputer, though there wasn't much missing data
Will you do cross-validation?
- Yes, we performed cross validation as part of our hyperparameter optimization
andytai7
Anupriya-Sri
2. Project set-up: Mechanics
Comments Where is the release?
3. Project proposal: reasoning
Comments What packages will you use for EDA what types of methods? What about class balance, there could be an imbalance in the classes, in which you would have to under sample or oversample. Which one will you utilize?
What about missing data, how will you handle the missing data?
For these algorithms, what packages will you use? Have you thought of using wrapper algorithms (boruta algorithm) for feature selection?
Will you do cross-validation?
A suggestion for metrics, to determine the performance of your models is Area Under Curve (AUC). The Area Under the Curve (AUC) is the measure of the ability of a classifier to distinguish between classes and is used as a summary of the ROC curve. Also look into and SHAP (SHapley Additive exPlanations) which explains the direction of each variable compared to the outcome variable.
5. Exploratory data analysis in a literate code document: VIZ
Comments I don't understand the graphs you created, you need an explanation for all of them, as well as a clarification of all the variables. IF this is not done the EDA, which is to inform of what variables are key, and some preliminary hypothesis is useless. In addition, EDA should be able to inform you on how to do some data wrangling and i don't see those conclusions.
5. Exploratory data analysis in a literate code document: REASONING
Comments Why are you using some of those EDA techniques? What's the purpose of a heat map etc.? Variables lists are useless unless you define what the variables represent. Needs higher resolution.