Sougata Roy

Dynamic Targeting serves as a strategic directive for field teams, leveraging the most current customer data and insights. Enhancement of this process is achieved through the consideration of the latest activities of HCPs, the application of Machine Learning (ML) algorithms, and the implementation of regular updates.

Key objective.

• Develop machine learning (ML) models first to identify similar characteristic HCPs. Subsequently, delve into assessing the impact of calls on sales. Later Identify the driver(s) which are contributing to the HCPs’ movement in Dynamic call plan compared to the usual call plan - either added or removed from call plan, or an increase or decrease in frequency of call.

Developed model to identify and prioritized certain specific type of vehicles for sales which are pending and taking high turnaround time with the salvage vendor.

Key objective:

• To Pre-process the data of more than 1 Million Records and hundreds of Features.

• To build and Train a Machine Learning Model using Logistics Regression, Random Forest Classification, Xgboost Classification.

• Involved in parameter tuning process for optimal model hyperparameters.

• To find the accuracy of the Model Prediction using Classification Reports, Confusion Matrix, AUC Score.

• Generate actionable insights from data and creating presentations to make recommendations for improvement The complete model builds in python environment

Active delinquent loans in a particular month can move to any three possible stages in next 12 months: Claim, natural cure and workout. The objective of the delinquent account status prediction model is to identify a loan moving into above mention three different segments in 12 months.

Key objectives:

• Multi nominal logistic regression technique to identify the loans moving to a particular segment with a definite prob. Value.

• The basic objective with this approach is to simultaneously all different probability of a loan moving to different segment in the next 12 months.

• Define the relationship between a categorical dependent variable and continuous/categorical independent variables.

• Saved $ value by implementing cost-saving initiatives that addressed long-standing problems.

US Mortgage Insurance organization seeks to establish an application forecast. Applications are steered by movement in macro- economic factor combine with internal policy decision. Business is looking to forecast monthly application volume for better resource planning.

Key objectives:

• A build a hybrid forecasting model, combining Time Series and ARIMA techniques.

• Forecasts of MI applications for next 7 months.

• Since macro-economic factors play a big role in steering this numbers, ARIMA technique use to predict monthly numbers, taking in account of effect of existing home sales, fixed rate interest, consumer confidence index.

• Weighted Average of TS component and ARIMA is applied to predict the final monthly numbers, where weights are decided by proportion of individual MAPE of these two-mention techniques.

To provide lenders with anticipated turn time of 4H and 9H application when they submit the data. Business was looking for growth rate driven by customer awareness and experience

Key Objectives:

• Build predictive model using three different in the process.

• Use robust regression Model to handle outlier's points.

• We have created hourly and weekday level matrix to check the variance.

• Also, we drill down analysis and identify at which level we are getting higher variance.

Provide quantitative and qualitative business insights to improved decision-making by gathering, analyzing and modelling client key performance indicators and market data using a broad set of analytical tools and techniques.

To interact with clients to understand the business need and identify the right set of predictor for the analysis.

To collate and present the business insights during client presentation.

US retail organization seeks to establish a case forecasting capability that will improve ability to accurately forecast cases delivered to each store per day to more efficiently plan labour allocation.

Key objectives:

• A build out of causal forecasting modelling infrastructure in SAS to support case forecasting for 400 US retail stores.

• Forecasts of cases delivered to each store per day by each network for next 98 days of US retail first quarter.

• Forecasts of trucks arrived to each store by day for RDC.

• I was involved in data preparation like (Data Validation and data cleaning) technique.

• We have used regression analysis to infer causal relationships between the independent and dependent variables.

• We have run PROC UCM, PROC VARMAX, PROC FORECAST, PROC TIMESERIES model in SAS

US pharma firm data management

Key Objectives:

• Generate analytical reports on weekly and monthly using SAS and SQL,.

• To review the status of data mapping and redundancy elimination.

• Pull out data, analysis and report stats inference as per client requests in SAS and SQL

Predictive Modelling

Key Objectives:

• To create a high-quality statistical-grade data set, emphasizing data validation, data cleansing & data reduction.

• Perform Variable reduction process starting with WOE and IV calculation and also perform PROC factor.

• Looking at IV value along with R^2 and (1-R^2) do variable section for final model.

• Use PROC LOGISTIC for prediction target variable and also perform statistical analysis of output file.

• Check the Null Hypothesis looking at H-L stats and do model validation on KS stats, lift and ROC curve.

Market Driver Model Analysis

Key Objectives:

• The objective was to identify the variables which drive the fund flow in the system.

• I was involved in data preparation like Data Validation, Missing Value Treatment, create month dummy variable and merge with main dataset.

• Prepare the Regression analysis to infer driver relationships between the independent and dependent variables.

• I was responsible for drafting Functional Specifications derived from client requirements and conveying these specifications to the software development and testing teams. When enhancements or modifications were necessary, I coordinated these updates with the relevant teams to ensure accurate implementation.
  
  Developed statistical models utilizing SAS and other analytical software tools.
• Engaged in data modeling with SAS, exploring alternative solutions to meet business requirements.
• Extracted data from raw data files using SAS import techniques and manipulated datasets utilizing SAS Data steps.
• Produced high-quality reports in various formats including listings, HTML, RTF, and PDF using SAS's Output Delivery System (ODS).
• Implemented new financial reporting standards (Schedule VI) and incorporated new financial ratios into the company database.
• Responsible for the preparation, review, and interpretation of financial statements, including income and cash flow statements, as well as balance sheets from previous company roles.
• Provided strategic decision-making support to an Indian client, analyzing stock price movements and other economic indicators such as interest rates, inflation rates, GDP.
• Developed robust statistical models using SAS by minimizing variability and simulating various scenarios to ensure model robustness.
• Conducted regular market surveys targeting customers in the finance industry to determine their requirements and satisfaction levels with our services.
• Created a Pivot tool for quick data summarization, analysis, and visualization in lists, tables, and pivot charts.
• Analyzed data using Pivot outputs, providing analytical recommendations to clients on improving database quality.
• Forecasted sales for the company and supported the research team with these insights.
• Participated in data preparation activities, including Data Validation, Missing Value Treatment, Time Series Analysis, Macro Economic Data integration, and developed statistical models using time series techniques like ARIMA.