European Statistics Awards – Nowcasting

This repository contains the replication code and reproducibility documentation for our submissions to the European Statistics Awards in Nowcasting.
The code reproduces the results that achieved several top placements across different challenges and indicators.

🏆 Awards and Results

🏖️ Tourism Nowcasting Challenge

• 🥇 Reproducibility Award (1st Place)
• 🥈 Accuracy Award (2nd Place)

Method: Seasonal-Trend Decomposition using Loess (STL)
The STL-based approach was used to nowcast the number of overnight stays, achieving high accuracy and robustness through decomposition of trend, seasonality, and remainder components.
Implementation relies on the stlm() function from the R forecast package.

📘 More details: European Tourism Nowcasting Challenge

---

🌍 Extra-EU Exports Nowcasting Challenge

• 🥈 Accuracy Award (2nd Place)

Method: ARIMA Modeling
An ARIMA-based approach was used to nowcast monthly Extra-EU export values. The model captures autocorrelation and seasonality effectively, balancing simplicity and predictive accuracy.

📘 More details: Nowcasting Extra-EU Exports Challenge

---

🌍 Intra-EU Exports Nowcasting Challenge

• 🥉 Accuracy Award (3rd Place)

Method: Seasonal-Trend Decomposition using Loess (STL)
The STL method was again applied for Intra-EU exports, demonstrating strong generalization and interpretability across multiple indicators.

📘 More details: Nowcasting Intra-EU Exports Challenge

---

📦 Repository Contents

• R/ — Model fitting and forecasting scripts (STL, ARIMA)
• reproducibility_reports/ — Documentation of steps and environment setup for full reproducibility

⚙️ Technical Summary

All analyses are implemented in R, leveraging the following key packages:

• forecast – STL decomposition and ARIMA modeling
• tidyverse – Data handling and visualization

---

👥 Team

Team bernese

• Marc Burri
• Sarah Schneeberger

---