Describe a complex academic or research problem you solved under tight deadlines. How did you break it down and what was the result?

Situation

During my undergraduate thesis, I built a predictive model for loan defaults using a dataset from a local microfinance NGO. Three weeks before the defense, I discovered that 30 percent of the repayment records had data entry inconsistencies, making early model runs inaccurate.

Task

I needed to clean and re-validate the data, re-run my logistic regression, and finalize the paper without asking for a deadline extension, while also preparing for the oral defense.

Action

I mapped out all variables and manually cross-checked a random sample of 200 entries against physical logbooks at the NGO office, then designed a set of validation rules to automatically flag discrepancies in the remaining data. I rebuilt the dataset in a structured Excel pipeline and re-ran the analysis, documenting every cleaning decision for transparency. While the model retrained, I drafted the methodology section and prepared supplementary exhibits for the panel.

Result

I submitted the manuscript on time and successfully defended it with a grade of 1.25. The final model achieved an 84 percent accuracy rate, and the NGOs operations manager later used my data-cleaning protocol to improve their internal reporting.