This project focuses on analyzing water treatment plant data using key parameters such as TDS (Total Dissolved Solids) and Turbidity. The goal is to evaluate system performance, detect anomalies, and generate actionable insights using data analytics and machine learning.
- Monitor water quality trends over time
- Evaluate treatment efficiency
- Detect anomalies in system performance
- Generate automated insights and reports
- Predict future system behavior
- Python
- Pandas (Data Processing)
- NumPy
- Matplotlib (Visualization)
- Scikit-learn (Machine Learning)
- Excel (Reporting)
-
Time Period: July 2025 – December 2025
-
Total Records: 67
-
Parameters:
- Raw TDS
- Treated TDS
- Raw Turbidity
- Treated Turbidity
- Loaded dataset using Pandas
- Converted Date column to datetime format
- Sorted data for time-series analysis
👉 Purpose: Ensures accurate chronological analysis
- TDS Efficiency
- Turbidity Efficiency
👉 Helps evaluate how effectively the treatment system removes contaminants
- Applied 5-day rolling average
👉 Purpose:
- Smooths fluctuations
- Reveals true trend
- Combined efficiency metrics into a single score
👉 Purpose:
- Provides overall system performance indicator
- Simplifies monitoring
-
Classified system into:
- Good
- Warning
- Critical
👉 Purpose:
- Mimics real-world monitoring systems
- Turbidity > 8
- TDS > 350
- Isolation Forest
👉 Purpose:
- Detect unusual system behavior
- Enable proactive maintenance
- Measured relationship between raw TDS and turbidity
👉 Insight:
- Strong correlation indicates common contamination sources
- Used Linear Regression to forecast future TDS
👉 Purpose:
- Anticipate system performance
- Support decision-making
-
Generated:
- Excel file
- CSV file
- Console summary
👉 Purpose:
- Deliver insights in usable formats
- Average TDS Efficiency ≈ 63.9%
- Average Turbidity Efficiency ≈ 88.1%
👉 Interpretation:
- Turbidity removal is highly effective
- TDS reduction is moderate
- Detected high TDS and turbidity values in early July
👉 Meaning:
- Possible filtration inefficiency
- Chemical imbalance
-
Detected both:
- High spikes
- Unusually low values
👉 Meaning:
- ML captures deeper patterns beyond thresholds
- Observed on specific days
👉 Possible Causes:
- Filter clogging
- Operational instability
- Strong positive relationship
👉 Insight:
- Same contamination source affecting both parameters
- Future TDS ≈ 239
👉 Meaning:
- System performance improving over time
- Shows raw vs treated TDS over time
👉 Insight:
- Clear reduction after treatment
- System effectiveness visible
- Raw vs treated turbidity
👉 Insight:
- Significant drop → efficient filtration
- TDS & Turbidity efficiency over time
👉 Insight:
- Stability and consistency of system
- Smooth trend of treated TDS
👉 Insight:
- Removes noise
- Shows actual performance trend
- Frequency of treated TDS values
👉 Insight:
- Identifies spread and outliers
- Evaluates consistency
- System performs consistently well for turbidity removal
- Moderate efficiency in TDS reduction
- Early-stage anomalies indicate operational issues
- Strong correlation suggests shared contamination source
- Predictive model indicates improving system trend
- Real-time monitoring dashboard (Streamlit)
- Advanced ML models for prediction
- Integration with IoT sensors
- Automated alert system (email/SMS)
Pranav More