This project analyzes fuel consumption patterns among drivers to identify potential anomalies in their fuel receipt claims. Using Z-scores, the analysis highlights data points where fuel consumption deviates significantly from expected patterns.
The goal of this project is to ensure accountability and detect potential issues such as fraudulent fuel claims, inefficient driving habits, or inaccurate mileage reporting.
Anomalies may occur under the following conditions:
Excessive Fuel Purchases: When the amount of fuel purchased is disproportionately high compared to the distance traveled. Underreported Mileage: When the kilometers reported are too low relative to the fuel purchased. Fuel Price Variations: When there are significant deviations in fuel cost that do not align with the base price per liter.
Data Input: Reads fuel consumption data from excel
Z-Score Calculation: Leverages pre-calculated Z-scores from the dataset to identify outliers.
Data Analysis:
- Metric Used: (Fuel (Liters) * Base Price) / KM
- Anomalies are flagged using Z-scores, with thresholds set at Z > 1.5 or Z < -1.5.
Visualizations:
- Scatter plots showing Z-scores over time.
- Anomalies highlighted based on Z-score thresholds.
- Threshold Customization: Users can adjust Z-score thresholds to redefine anomaly detection
KH Driver.csv: Contains the fuel cost data, including the calculated Z-scores.
