Projects

Below you may find my Data Science Projects and Other Resources.

Click on the links to download the file of interest.

Project	Content
Production Function Estimation and Analysis (Python Jupyter Notebook)	Topics: Production Function Estimation, Firm Productivity Analysis, COGS and SG&A Elasticities, Industry-level Economic Modeling. – Estimated firm-level production functions using Compustat financial data, focusing on a Cobb–Douglas specification linking output to short-run inputs (COGS) and long-run inputs (capital). – Implemented the De Loecker, Eeckhout, & Unger (2020) methodology to compute industry-level input elasticities over time through log-linearized regressions estimated sequentially by year. – Extended the baseline model to incorporate SG&A expenses as an additional input in the production function framework. – Developed Python code to clean and structure panel data, run regressions, and estimate input elasticities and firm-level productivity measures using GMM.
Job Postings Newsletter (PDF) Job Postings Analysis (Python Jupyter Notebook)	Job Postings Data Wrangling and Analysis of Macroeconomic Trends (Notebook and pdf with newsletter). Topics: Macroeconomic Trend Analysis, Industry and Occupation Trend Decomposition, Time-Series Construction and Quarterly Aggregation of High-Frequency Data. – Processed and combined millions of LinkedIn job posting records, including firm identifiers, role categories, salary information, posting/removal dates, and industry classifications. – Cleaned, enriched, and aggregated high-dimensional microdata (e.g., by quarter and industry) to produce summary statistics such as vacancy counts, average salaries, and posting durations. – Analyzed trends in job vacancies across industries and geographic areas, highlighting periods of expansion and contraction in labor demand. – Integrated external macroeconomic indicators (e.g., unemployment rates and GDP growth from FRED) to examine correlations between labor market activity and broader economic conditions over time.
Feature Selection, Mortgage Approval Predictive Model (R Markdown)	Topics: Mortgage origination, HMDA data, Logistic regression, Model comparison, Regularization (Ridge/Lasso), Predictive performance. – Cleaned and structured detailed HMDA loan-level data, including loan terms, borrower characteristics, and property attributes. – Developed and compared multiple predictive model specifications to estimate the probability of loan origination. – Evaluated the impact of alternative predictor sets (e.g., interest rates, debt-to-income ratios, property values) on model performance. – Assessed model accuracy using performance metrics such as AUC, confusion matrices, and overall classification accuracy to identify best-performing specifications.
Hospital System Affiliation and Financial Performance: Event-Study Evidence from Multiple Specifications (Python Jupyter Notebook)	Topics: Hospital System Entry, Difference-in-Differences, Event Study, Hospital Financial Outcomes, Robustness Checks, Causal Inference. – Cleaned and structured panel data on California hospitals, including annual revenues, expenses, and financial margins. – Constructed treatment indicators capturing the timing of hospital system affiliation. – Implemented a difference-in-differences event-study design to estimate dynamic effects of system entry on financial outcomes. – Analyzed changes in net income and operating margins before and after affiliation, documenting shifts in financial performance around system entry.

Resource	Content
Study Notes in Macroeconomic Theory	Topics: consumption under uncertainty, real business cycles, asset pricing, incomplete markets, and limited commitment.
Study Notes in Stochastic Macroeconomic Theory	Topics: Value Function Iteration, Log-linearization, and the perturbation method to compute the Impulse Response Functions.
Solving DSGE and DGE Matlab Codes	Topics: Value Function Iteration, Log-linearization and the perturbation method to compute the Impulse Response Functions.

Disclaimer: the views presented on this website do not reflect the opinions of Analysis Group or its work. All errors are mine.