Modeling & Causal Inference
From systems to human behavior
Built flooding/traffic/queue simulations and applied Synthetic Control to estimate a +12.6% policy effect on fertility.
Synthetic Control for Fertility
Constructed donor pool; identified 12.6% post-reform effect.
We studied the 1994 French Parental Leave Reform using panel data and synthetic control to address the causal question directly. By 2001, the reform increased total fertility by 12.6% – about +0.24 children per mother. I co-designed the study, selected countries for the donor pool, and compiled the dataset with two collaborators, emphasizing pre-treatment fit and robustness checks to support identification.
Methods: synthetic control, DID intuition, panel data, causal inference, robustness checks
Role & context: Team of 3; SS154 Econometrics (senior year), Minerva
Abstract France serves as a case study for how to successfully improve a country’s total fertility rate. In 1963 France’s fertility rate peaked at 2.89, which was followed by a significant decrease to the lowest point in the country’s history at 1.73 in 1994. In the same year, the government reformed the parental leave system, leading to a steep increase in the total fertility rate, so this paper raises the causal question of ”What was the causal effect of the 1994 French Parental Leave Reform on France’s total fertility rate?”. Using the synthetic control method, it is estimated that the new policy increased France’s fertility rate by 12.6% evaluated in the year of 2001 - it motivated people to have 0.24 more children on average per mother.
Cellular Automata Flooding Model
Cellular-automata flooding model and policy scenario testing.
I built a cellular-automata simulation over GIS-style grids to forecast flood spread and depth under varying rainfall and terrain assumptions. I presented the model for a high-risk region in Bulgaria to the head of the Bulgarian Academy of Sciences’ climate institute. The workflow emphasized the need for data and prevention policies and actions.
Methods: cellular automata, GIS-style grids, hazard modeling, policy scenario testing
Role & context: CS166: Modeling and Analysis of Complex Systems course; presented to Institute head (Bulgarian Academy of Sciences)
Hyderabad Traffic Micro-Simulation
Cellular automata to optimize road throughput.
I implemented a cellular-automata traffic model for a major road in Kondapur, Hyderabad, motivated by sustained, non-stopping flow, nearby where I lived at the time. I evaluated lane and signal configurations to relieve bottlenecks, using sensitivity analysis to test robustness across demand levels and driver behaviors. The model supports data-driven design decisions by quantifying throughput, delay, and density under alternative layouts.
Methods: cellular automata, micro-simulation, throughput optimization, sensitivity analysis
Role & context: CS166: Modeling and Analysis of Complex Systems course; major road in Kondapur, Hyderabad
Grocery Checkout Queue Design
Monte Carlo and M/G/1n simulation balancing wait time and staffing.*
We applied queuing theory to determine the optimal number of cashiers in a supermarket for a hypothetical scenario. Using an M/G/1 * n formulation with Monte Carlo simulation, I quantified trade-offs between expected queue length, service time, and staffing cost under realistic arrival/service distributions. The result is a policy that meets target waits while minimizing over-staffing risk.
Methods: queuing theory, M/G/1 * n, Monte Carlo simulation, operations research
Role & context: CS166: Modeling and Analysis of Complex Systems course; operations research module
