Projects

Predicting Currency Crises in Emerging Markets: A CART-Based Early Warning System

This study applies machine learning (CART) to predict currency crises in five emerging markets, finding that real exchange rate overvaluation is the strongest universal predictor, while country-specific vulnerabilities vary significantly — from Argentina's current account deficits to Turkey's short-term debt exposure.

Price Discrimination Strategies of Air India and Vistara on the Delhi-Mumbai Route

An analysis of Air India and Vistara's pricing behavior on the busy Delhi–Mumbai route, focusing on how different forms of price discrimination are applied and what they reveal about competition, demand patterns, and airline revenue strategies.

The Taylor Rule in Brazil: An Empirical Analysis of Monetary Policy

This paper examines the application of the Taylor Rule to Brazil's monetary policy from 2003 to 2023, comparing both basic and expectation-based formulations. Using output gap calculations based on the production function approach and inflation targeting data, we analyze how well the Central Bank of Brazil's SELIC rate aligns with Taylor Rule prescriptions and evaluate the rule's effectiveness in an emerging market context.

Convergence analysis [work in progress]

Testing the convergence hypothesis in economics and exploring whether countries' incomes and productivity levels move closer together over time.

Flight Analysis

Exploratory data and machine learning analysis of flight delays and prediction patterns using commercial flight data.

Traffic prediction on PEMS dataset using a Spatial-Temporal Dependent Convolutional LSTM model

Currently implementing a novel deep learning approach for traffic prediction using the PEMS (Performance Measurement System) dataset. This research implements a Spatial-Temporal Dependent Convolutional LSTM (STD-ConvLSTM) model that captures both spatial dependencies between traffic sensors and temporal patterns in traffic flow.

Atharva Kulkarni