• Performed Marketing Mix Modelling (MMM) to understand how marketing activities are driving the profatibility of multiple products of a CPG company and how to allocate their marketing budget.• Identified the key sales drivers (marketing media, price, competition, covid impact on sales and other macroeconomic factors).• Simulated S-curves to describe, visualize and predict business performance progressively over time,with an aim to increase ROI.

Supervisors: Dr Kiran Garimella (IDSS, MIT); Prof Aaditya Dar (ISB Hyderabad); Ms Vasundhara Sirnate (The Polis Project)• Built a text classifier to identify incidents of violence in news articles over past 20 years from The Times of India.• Did topic modelling to discover the different topics covered in the news and analyse their trends over time. Also used the same to filter out topics related to non-violence and thus reduce the degree of imbalance in the data.• Used Named Entity Recognition(NER) to identify the number of fatalities involved in the cases of violent incidents.

• Dataset related to attributes of different kinds of pollutants that were collected from several industrial sites across the city.• Analyzed patterns in data and applied hierarchical clustering on them on the basis of pollution level, similar site codes, etc.• Used wavelet spectral analysis on time series data to detect and analyze peak values, irregularities and possible outliers.

• A task based on the balanced binary classification of congestion type in the telecommunication network of a city.• Used Recursive Feature Elimination to identify which attributes contribute the most towards predicting the target variable.• Achieved an accuracy of 91% on unseen data using the final trained model based on XGBoost tuned with gridsearchCV .