Handling imbalanced data using Oversampling methods(SMOTE, ADASYN, and DB-SMOTE)

Using Accept-Reject method a new Beta R.V. generated

Random variables, and generating random variables using inverse transform method

Implemented the following root finding methods: Interval Bisection Method, Newton-Raphson Method and Secant Method

Quasi Poisson logistic regression, Extra Poisson variation, Negative Binomial logistic Regression and Drop in deviance test to compare the models.

Poisson logistic regression (Null to Saturated), Lack of fit test and Drop in deviance test to compare the models

Proportional odds logistic (Ordinal logistic) regression, Drop-in-deviance to compare the models, finding probabilities and comparing with Multinomial logistic regression (Ignoring order)

- Multinomial Logistic Regression (Saturated to Null) - Drop-in-deviance to compare the models - Finding probabilities for the special cases

Model comparison (saturated to null), GLM models with binomial and quasibinomial, analyzing Overdispersion.

Compared GLM models with full(saturated) model and performed Drop-In-Deviance test to choose the best model

Testing Mutual, Joint and Marginal independences using Pearson’s Chi-Squared test statistics.

Joint and conditional distributions for two categorical variables, creating a contingency table and hypothesis test for independence and Odds Ratio & Relative Risk.

Odds Ratio, Relative Risk, Hypothesis test for independence and confidence interval

Joint and conditional distributions for two categorical variables, creating a contingency table and test for independence

Simulating expected value of discrete random variables

Using Discrete random variables (Bernoulli, Binomial, Geometric, Hyper Geometric and Poisson) solving probability questions.

Solving probability problems by simulating in R

Accuracy, Recall, Precision and F1 scores are compared for balanced and imbalanced set sets

Following performance metrics were evaluated: Accuracy, Precision, Sensitivity, Specificity, and F1 Scores

ROC Curve, AUC and choosing the best cut off value

Performed SVM using linear, radial and polynomial kernel and tuning to choose the best cost parameter

Performed Decision tree and Pruned tree through Cross Validation . Used Rpart library

R- Resampling methods

R- Classifications

R- Data Wrangling in the Tidyverse

R- Subsetting Objects

R- Basic Exploration

Performed Decision tree, Bagging, Boosting, Feature selection and Random forest on Carseats daat set

Used Ridge, Lasso, Principal Component and Partial Least Squares regression techniques

Performed subset selection using AdjR2, Cp and BIC, Forward and Backward stepwise selection, Lasso regression

Estimate the test error rate of Naive Bayes model using data split, LOOCV, K-Fold CV, Repeated K-Fold CV and Bootstrap methods.

Logistic Regression, LDA, QDA and KNN methods were compared on Auto data set

Logistic Regression, LDA, QDA & KNN models

Used Backward-stepwise, Bootstrapping and VIF for feature selection