Apply Ideas¶
Technique: Create derived metrics (signals) that reveal system behavior more clearly than raw measurements.
Signals help transform raw observations into interpretable indicators.
A good dataset for this module:
- contains multiple related measurements
- allows ratios, averages, or rates to be calculated
Example Systems¶
Website Traffic¶
Possible fields:
- requests
- errors
- total_latency_ms
Possible signals:
- error_rate = errors / requests
- avg_latency = total_latency_ms / requests
Questions to explore:
- Which signals best indicate system health?
- Are raw counts or derived signals easier to interpret?
Retail Sales¶
Possible fields:
- units_sold
- revenue
- customers
Possible signals:
- revenue_per_customer
- units_per_customer
Questions to explore:
- What signals indicate strong or weak performance?
- Which signals help compare days with different traffic levels?
Fitness Tracker Data¶
Possible fields:
- steps
- active_minutes
- calories
Possible signals:
- calories_per_minute
- steps_per_minute
Questions to explore:
- Which signals reveal intensity of activity?
- Do derived metrics change how behavior appears?
Environmental Monitoring¶
Possible fields:
- temperature
- humidity
- wind_speed
Possible signals:
- temperature_change_rate
- comfort_index
Questions to explore:
- What signals help interpret environmental conditions?
- Are some signals easier to interpret than raw values?