The zip file contains the following files:
# IMPORTANT : When using logistic_regression, reg_logistic_regression, logistic_regression_newton, reg_logistic_regression_newton, the initial weight vector should be of shape (n,) and NOT (n,1)

*1* run.py : This is the main file to be run. As it is, it will perform regularized logistic regression using Newton's method. 
The basic outline of the file is as follows:
	- Read the training and test data
	- Split both sets of data (test and training) by the jet numbers 0 - 3. Split all the jets further by valid or invalid DER_MASS_MMC. For each jet, remove the features that are either
		not calculated (-999) or deemed irrelevant (See [1]).
	- Loop through each data set (8 of them, 2 for each jet - 1 with valid DER_MASS_MMC and the other with invalid DER_MASS_MMC)
		- Standardize feature set after removing outliers above a threshold specified in list_sd_limit
		- Conduct a polynomial expansion of the features (degree 2)
		- Obtain the weights and final loss function by using the desired classification model (currently logistic regression with Newton's method and lambda = 1.
		- Standardize test feature set using the mean and standard deviation of the training data set.
		- Get the predictions for the test data set for the current jet and mass validity. 
		- Add the predictions to the overall prediction list
	- Print the predictions to file, with the ids and corresponding predictions

*2* proj1_helpers.py - 
	- Standardization and outlier removal
	- Read, write data
	- Obtain and verify predictions
	- Split data by jet numbers + removal of unnecessary features
	- Polynomial expansion to the specified degree
	- Split data into test and training for cross validation
	
*3* implementations.py : All the regression models required
	- Linear:
		- Normal equations
		- Ridge regression
		- Gradient Descent
		- Stochastic Gradient Descent
	- Logistic
		- Gradient Descent
		- Regularized Gradient Descent
		- Newton's 
		- Regularized Newton's
	- Other functions required to calculate gradients, the log function and the Hessian

* ----------------------------------------------------------------------------------------------------------------------- *	
[1] Features removed for each jet number

Jet 0, valid DER_MASS_MMC   --> [   4, 5, 6, 8, 12, 15, 18, 20, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32] 
Jet 1, valid DER_MASS_MMC   --> [   4, 5, 6,    12, 15, 18, 20, 22,         25, 26, 27, 28, 29,     31, 32]
Jet 2, valid DER_MASS_MMC   --> [                   15, 18, 20, 22,         25,         28                ]
Jet 4, valid DER_MASS_MMC   --> [                   15, 18, 20, 22,         25,     27,     29            ]
	
Jet 0, invalid DER_MASS_MMC --> [0, 4, 5, 6, 8, 12, 15, 18, 20, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32]
Jet 1, invalid DER_MASS_MMC --> [0, 4, 5, 6,    12, 15, 18, 20, 22,         25, 26, 27, 28, 29,     31, 32]
Jet 2, invalid DER_MASS_MMC --> [0,                 15, 18, 20, 22,         25,         28                ]
Jet 3, invalid DER_MASS_MMC --> [0,                 15, 18, 20, 22,         25,     27,     29            ]

0	DER_mass_MMC			15	PRI_tau_phi					30 CUSTOM_assymenergy
4	DER_deltaeta_jet_jet	18	PRI_lep_phi					31 CUSTOM_delta_phi
5	DER_mass_jet_jet		20	PRI_met_phi					32 CUSTOM_avg_phi
6	DER_prodeta_jet_jet		22	PRI_jet_num					33 CUSTOM_special 
8	DER_pt_tot				23	PRI_jet_leading_pt
12	DER_lep_eta_centrality	24	PRI_jet_leading_eta
							25	PRI_jet_leading_phi
							26	PRI_jet_subleading_pt
							27	PRI_jet_subleading_eta
							28	PRI_jet_subleading_phi
							29	PRI_jet_all_pt