Plan

Plan LAST WEEK v1.0
Send an email to Chemicum!
1. "Out of bag" prediction in python: last search                        ALINA
2. Improve the architecture of results:                                  ALINA
   2.1. How to get automatically evaluation of the "error"?
2. Feature selection: develop algorithms to apply to SVM notebook        ALEKSANDRA
['AMR', 'VP-3', 'SsOH']
3. Update SVM notebook:                                                  PRASOON
   3.1. Run models of all kernels
   3.2. Try playing with other parameters
4. Start developing notebook for other classifiers in our plan...


Plan week Dec 2-8
1. Finish RF models:                                                     ALINA
   1.1. Imbalanced towards B in 4:1 ratio ("oversampling")
   1.2. Get "out of bag" prediction printed for all 3 models
   1.3. Calculate differences ("standard errors") for RF models
2. Feature selection:                                                    ALEKSANDRA
   2.1. See sklearn.feature_selection functions
        See M5_report Task 3 for details
3. Create a notebook for SVM algorithm:                                  PRASOON
   3.1.Make Train-test split on Train data (don't use validation set yet)
   3.2.Run nB and balanced models
        using 3 original features
		try linear kernel and more
		comment the notebook


Plan week November 18-24 --> gets postponed because of HW10 on the project
1) oversampling for Model3: Imbalanced towards B compounds           PRASOON --> see publication
2) How to get model metrics on train set? - try '.predict'(X_train)' ALINA --> not correct approach
3) How to to feature selection?                                      ALEKSANDRA (postponed)
4) Run SVM with 3 original features (create a notebook...)           ALINA will start (postponed)


Plan 08/11/19
Prasoon: Read through the materials
Aleksandra: Cleaning the data (see 2.1 in the publication; check the web link of publication and splitting the data) +
			Split the training set from the paper into training2 and validation2
			Go through SVM homework materials
Alina: Understand model in the paper
Support Vector Machine