Pooja M P

Data Analyst with hands-on expertise in Power BI, SQL, Python, and R, and a Master's in Agricultural Statistics. Built and validated predictive models that cut forecasting error by up to 80.6% over baseline statistical methods, and designed interactive Power BI dashboards with dynamic DAX measures for workforce and sales analytics. Strong foundation in hypothesis testing, regression, and time-series forecasting, with certifications from Google and Samsung.

Looking to bring rigorous, data-driven analysis to Edwagon's Data Analyst team.

Graduate Researcher — Weather-Based Forecasting of Mango Powdery Mildew - University of Agricultural Sciences, Bangalore - Bangalore, India

(2024-01 - 2026-01)

• Forecasted mango powdery mildew severity with 68.9% lower prediction error (RMSE 2.26 vs. 7.26) than the best statistical baseline by building and validating a Support Vector Regression model on 8 years (2017–2025) of disease and climate data across 12 engineered features.
• Identified significant drivers of disease severity (p < 0.0001) — maximum temperature (r = 0.757) and evening relative humidity (r = -0.850) — using two-way ANOVA, Tukey's HSD, and Pearson correlation, directly informing feature selection for downstream models.
• Benchmarked 3 statistical models (MLR, ARIMAX, GAM) against 5 ML models (Decision Tree, Random Forest, XGBoost, ANN, SVM); SVM outperformed all others while holding near-identical RMSE across training, validation, and testing, confirming minimal overfitting relative to XGBoost and ANN.
• Delivered publication-quality analysis and visualizations summarizing an 8-year Percent Disease Index dataset, supporting future field-level disease-management decisions.

Master of Agricultural Statistics - Agricultural Statistics - University of Agricultural Sciences, Bangalore (2024-01 - 2026-01)

Bachelor of Science (Hons.) - Agriculture - University of Agricultural Sciences, Bangalore (2024-01)