This project analyzes ORU’s off-season sewer usage using Python, with pandas for data handling, histograms and line plots for exploration, and a scipy-based model for prediction. Pearson’s correlation and visualizations help reveal key trends and relationships.
Executive Summary
The project aims to understand and reduce sewer usage during the off-season (November to April) at ORU. It explores factors influencing sewer usage, including cooling degrees and water usage, and builds a predictive model to minimize off-season usage.
Goals of the Project
Understand Off-Season Sewer Usage: Analyze factors affecting sewer usage during off-season months and hypothesize methods to reduce it. Analyze Relationships: Examine how cooling degrees, water usage, and sewer charges relate to sewer usage. Predictive Modeling: Develop a model to predict and reduce off-season sewer usage. What We Did
Exploratory Data Analysis: Generated histograms and line plots to capture usage patterns and seasonality. Usage Comparison: Compared off-season and on-season sewer usage to identify high-contributing meters. Predictive Modeling: Created a model with a low Mean Squared Error to estimate future sewer usage. Results
Usage Patterns: Histograms revealed significant variations in water and sewer usage, highlighting seasonal trends and anomalies. Seasonal Trends: The summary line graph showed a decrease in sewer usage from 2012 to present, with a spike in off-season usage from 2014 to 2017. Trouble Meters: Identified meters with higher off-season usage, such as Quad.Maintenance and Hamill_Timko_Dorms, which are key targets for intervention. Usage Comparisons: Significant decreases in water usage during the off-season were noted, with the NEC, Welcome.Center, and Chapel being less impactful compared to the major users. Yearly Trends: Off-season sewer usage generally exceeds on-season usage from 2013 onwards, with notable peaks and fluctuations. Discussion
The analysis provides valuable insights into sewer and water usage patterns, identifying key meters contributing to off-season usage. The predictive model shows promise, but further validation is needed.