# An evaluation of time series forecasting models on water consumption data: A case study of Greece

**Authors:** Ioannis Kontopoulos, Antonios Makris, Konstantinos Tserpes, Theodora Varvarigou  
**arXiv:** [2303.17617](https://arxiv.org/abs/2303.17617)  
**Date:** 2023-03-30  
**Summary page:** https://www.summarizepaper.com/en/paper/2303.17617v1/

## Key points

- Rapid urbanization and industrialization have led to an increasing demand for water resources, creating a significant gap between supply and demand.
- Proper water distribution and forecasting of consumption are crucial factors in mitigating the imbalance by improving operations, planning, and management of water resources.
- Ioannis Kontopoulos et al. evaluate several well-known forecasting algorithms over time series data from Greece to reveal key insights about each algorithm's effectiveness in predicting water consumption patterns.
- The study evaluates various forecasting algorithms such as ARIMA, Exponential Smoothing, Seasonal Naïve Bayes, Artificial Neural Networks (ANN), Support Vector Regression (SVR), Random Forests (RF), Gradient Boosting Machines (GBM), and Long Short-Term Memory Networks (LSTM).
- The results show that LSTM outperforms all other models in terms of accuracy in predicting future water consumption patterns.
- Socio-economic factors such as population density, income levels, weather conditions significantly impact water usage patterns.
- The findings can inform policymakers' decisions on how best to manage their water resources effectively while meeting the growing demand for clean drinking water.

## Summary

The rapid urbanization and industrialization of recent years have led to an increasing demand for water resources, creating a significant gap between supply and demand. To address this issue, proper water distribution and forecasting of consumption are crucial factors in mitigating the imbalance by improving operations, planning, and management of water resources. In this paper titled "An evaluation of time series forecasting models on water consumption data: A case study of Greece," Ioannis Kontopoulos et al. evaluate several well-known forecasting algorithms over time series data from Greece, a country with diverse socio-economic and urbanization issues. The authors analyze a real-world dataset provided by the Water Supply and Sewerage Company of Greece to reveal key insights about each algorithm's effectiveness in predicting water consumption patterns. The study evaluates various forecasting algorithms such as ARIMA, Exponential Smoothing, Seasonal Naïve Bayes, Artificial Neural Networks (ANN), Support Vector Regression (SVR), Random Forests (RF), Gradient Boosting Machines (GBM), and Long Short-Term Memory Networks (LSTM). The results show that LSTM outperforms all other models in terms of accuracy in predicting future water consumption patterns. Additionally, the authors find that socio-economic factors such as population density, income levels, weather conditions significantly impact water usage patterns. Overall, this study provides valuable insights into the effectiveness of various forecasting algorithms for predicting water consumption patterns in countries facing diverse socio-economic challenges. The findings can inform policymakers' decisions on how best to manage their water resources effectively while meeting the growing demand for clean drinking water.