The project was based on building full process on Azure database and RStudio workbench. The project consisted on buidling Shiny Apps to validate data through clicks and run models on server on real time basis and push the output on Shiny App to create a dynamic dashboard consisting of various visualization and business interpretation as per client requirements. Packages with unit testing and Shiny Apps was delivered to be used for Global Production

Worked on automating all the MIS production for monthly as well as quarterly business reporting to stakeholder increasing efficiency. The codes in SAS and Excel were converged to make the outputs get generated within limited time.

Worked on building automated GLM model for identification of different factors impacting the business leading to higher claims and low premium rate. The factors are hence used for adjustment in the pricing model to get better rates for premium calculation.

Using longitudinal data to build complete model and sub-models based on clustering of data to find sub-groups. The thesis involves use of following techniques: GLM, Lasso and Ridge Regression Clustering of longitudinal data using Euclidean distance based on space and time Building sub-models based on the clustering Building classifier using SVM and Decision Trees After forecasting, categorization done to see the status of carbon emission and change in it

Using fuzzy matching techniques to build machine learning models to compare records in large amount of data with focus on making the algorithm efficient enough to run in short time. Calculating string distance using different methods and giving weightage to each variable to calculate record measure for comparison. Using Machine Learning techniques to recalculate the weights for effective comparison and reduce the runtime of large amount of data.

Worked on the testing and modification of the tool developed by Milliman. The task was to test the working of the premium and claims code developed by Milliman and aligning it with the process developed by Analytics team. Testing the reinsurance model and finally the output moving into creation of triangles for further calculations and reconciliation.

Worked on a project related to Financial Markets wherein I had to examine the trust-worthiness of a prospective customer and his/her possibility of defaulting on loan. The task was to build a Behavioural Credit Risk Model based on a large sample of data by applying Logistic Regression on SAS and R The data had various information related to behaviour transactional details of the customers and their performance in repayment.

Worked on a project to predict the customer churn and retention rates in telecom industry using Markov Chain Questionnaire was prepared to collect data regarding change in the network within a time period The count of shift was then converted to transition probability matrix and was checked for stationarity Analysis was done based on the results obtained