Time series forecasting of Dissolved Oxygen (O₂) in the Southern Bug River using ARIMA, SARIMA, Holt-Winters, and Gradient Boosting on 21 years of water quality data. Best model: Holt-Winters with MAPE = 9.01%.

Using prophet forecasting to predict individual household power consumption

Time series forecasting of NO₂ pollution in London (2010–2023) using SARIMA and Prophet. Includes baselines, log-transforms, ensemble methods, and horizon sensitivity analysis.

A comprehensive time series forecasting project applying both classical statistical models (ARIMA) and deep learning methods (LSTM) to Johnson & Johnson EPS and Amazon stock price data. Includes exploratory analysis, seasonal decomposition, ADF testing, model tuning, multi-step forecasting, and FFT-based frequency analysis.

Eexploration of deterministic processes in time series forecasting using statsmodels, featuring trend modeling, seasonality, Fourier terms, custom components, and integration with AutoReg and SARIMAX models.

Forecasting short-term household energy consumption using ARIMA, Facebook Prophet, and XGBoost — with interactive Plotly visualizations, advanced feature engineering, and comprehensive model evaluation on the UCI Household Power Consumption dataset.

Zeitreihenanalyse & Vorhersage des Wasserstands der Reuss bei Andermatt (CAMELS-CH) — Zerlegung, Feature-Auswahl & Extremwertanalyse

End-to-end big data analytics pipeline on 500K+ Google Maps business reviews using PySpark, NLP, ALS Collaborative Filtering (RMSE 0.9896), ARIMA time series forecasting, and seasonal decomposition — delivering actionable business intelligence on customer behavior, review trends, and personalized recommendations.

Zero-Inflated Seasonal Energy Consumption Forecasting

Analyze and forecast natural gas prices using time series data, with seasonality decomposition and signal detection for trading strategy insights.

Analyzed U.S. environmental emissions and industrial data using Tableau and Power BI, focusing on climate change impacts. Developed dashboards to visualize trends, cleaned data with Tableau Prep, and highlighted industry-specific and regional variations to guide emission reduction strategies

A taxi company called Sweet Lift has collected historical data on taxi orders at the airport and they need to predict the number of taxi orders for the next hour.

Hourly taxi-order forecasting for Sweet Lift Taxi using LightGBM with lag features and rolling means (RMSE 39.7, target ≤ 48).

A Python tool for seasonal decomposition of time-series CSV data, with reproducible preprocessing and visualization.

A comprehensive implementation of the Theta Model for time series forecasting. Includes model estimation, sensitivity analysis, diagnostics, and extensions. Uses statsmodels, pandas, and matplotlib.

End-to-end time series analysis of hourly electricity consumption across 4,400+ London households using smart meter data