residual-diagnostics

Temporally partitioned wind-power forecasting on the DaKS/Kassel dataset, with full-cohort tabular benchmarks, four-park neural/sequence experiments, and validation-calibrated residual screening for PHM-oriented research.

Project work for Time Series Analysis. Includes exploratory analysis, ARIMA modeling, diagnostics, forecasting, and evaluation using R. Covers trend/seasonality modeling, stationarity checks, ACF/PACF analysis, model selection, and forecast accuracy assessment.

STL decomposition notebook with parameter analysis, diagnostics, seasonal strength metrics, forecasting, and alternative methods. Complete time series decomposition workflow with visualizations and statistical tests.

Built and validated a linear regression model to predict annual customer spend using session and membership features; achieved R² = 0.981 and RMSE ≈ $10 on hold-out test data.

A comparative audit of 15+ time series architectures (ARIMA-GARCH, XGBoost, Transformers) across four temporal domains. Focuses on predictive robustness, residual adequacy (Ljung-Box, ARCH-LM), and statistical significance testing.

Regression analysis of climate variables and global crop yields using R

PIT-based quantile and randomized quantile residual (RQR) diagnostics for Stata models.

Four-stage residual-diagnostic extension of the Creator Variable Machine (Shishegaran & Varaee) for detecting missing physical variables — validated on synthetic systems and RC deep beam shear capacity.

Time-series modelling of open-ocean temperature anomalies using stationarity testing, ARMA model comparison, residual diagnostics and holdout forecasting in R.

R and Python diagnostic tools for linear mixed models: indexed recursive residuals, CUSUM, graphical model checking, and R bootstrap source.

Does experience or skillset predict developer salary more? SQL + statistical modeling on 4,600+ real survey responses.