overall dashboard showing score trends and predicted risk
overview
contextI build decision-focused tools from public inspection data. The goal is to make risk easier to interpret and help prioritize limited inspection resources.
- Dashboard: separates current performance vs predicted decline to reduce misinterpretation.
- ML (in progress): building toward an outcome-risk model using public inspection records.
- Focus: labeling, readability, and fast interpretation for non-technical users.
tools
stack
power bi
dax
data modeling
geospatial viz
python
pandas
scikit-learn
Data source: Mecklenburg County restaurant inspection records.
Food Establishment Risk Dashboard
power bi • daxDashboard built from Mecklenburg County inspection records. It separates current inspection performance from predicted decline to keep decisions clear.
- Custom DAX measures combine recent scores and trends into a risk indicator.
- Geospatial mapping highlights clusters of higher-risk establishments.
- Labels/colors were adjusted based on feedback to improve readability.
map + table used to identify establishments needing earlier inspection
tooltip details showing how to interpret specific locations
Predicting Inspection Outcomes (In Progress)
python • mlMachine learning project using public inspection records to estimate compliance risk. Current focus is data preparation and feature engineering.
- Preparing inspection + violation history data for modeling.
- Engineering features from inspection patterns and establishment type.
- Evaluating classification approaches for higher-risk detection.
- Comparing results to published methods.
classification
feature engineering
data cleaning
evaluation