Envelopa Mobility | Other Findings

What influenced mobility at the Envelopa campus?

This page brings together four findings from the statistical analysis behind the dashboard. It shows how mobility related to weather, whether the academic calendar improved the explanation of daily volumes, and how often unusual daily values occurred.

Choose a topic below, hover over the chart bars for exact values, and use the interpretation panel to understand what the result means in practical terms.

Which modes were most sensitive to weather?

Average strength of the relationship with temperature and precipitation.

Hover over a bar to view the exact correlation and number of tests.

Active travel responded more strongly

Cycling and walking were more closely associated with weather conditions than motor-vehicle flows.

Temperature was the clearer influence for all three modes. Pedestrian and cyclist counts changed more consistently with temperature than with precipitation, while motor-vehicle counts remained comparatively stable.

Cycling showed the strongest single relationship in the analysis, reaching an absolute correlation of 0.744. This does not mean every cycling location reacted equally, but it shows that some bicycle observations were particularly weather-sensitive.

CyclistsMean |ρ| 0.305 · 6 tests

PedestriansMean |ρ| 0.296 · 4 tests

Motor vehiclesMean |ρ| 0.117 · 20 tests

How did mobility change during specific weather conditions?

Median adjusted change in daily mobility volume during cold, warm and precipitation days.

Hover over a bar to view median change, mean change and significance.

Cold weather had the clearest effect

Cycling fell most strongly on cold days, with a median adjusted decrease of 26.82%.

Cold conditions produced a larger and more consistent response than precipitation or warm days. Every cyclist cold-day test was statistically significant, while the pedestrian decrease was smaller and not significant.

Precipitation was also associated with lower cycling volumes, but the result was less consistent across tests. Motor-vehicle changes were small under all three weather conditions.

Cyclists · cold days−26.82%

Cyclists · precipitation−12.97%

Motor vehicles · warm days−3.55%

Pedestrian warm-day effects were not tested because the Telraam observation period did not include any warm days under the analysis definition.

How common were unusual daily observations?

Share of daily observations shortlisted as deviations within each dataset.

Hover over a bar to view the deviation count and number of unique days.

Deviations were uncommon and rarely shared

Fewer than one in ten observations were shortlisted as deviations in every dataset.

Only a minority of unusual dates matched known local events or Czech public holidays. This means the deviation shortlist should not be interpreted as a simple calendar of event impacts.

No date appeared as a deviation in all three datasets. The limited overlap indicates that many unusual values were more likely related to a particular mode, source or monitoring location.

EnCLOD70 days · 20 accounted for

Telraam25 days · 6 accounted for

Nextbike11 days · 1 accounted for

All three datasets0

EnCLOD + Telraam12

EnCLOD + Nextbike4

Telraam + Nextbike0

Did the academic calendar help explain daily mobility?

Improvement after academic term was added to models already accounting for weather and weekday.

Hover over a bar to view the median ΔAIC, test count and improvement rate.

The university calendar added useful context

Academic term improved nearly every model, even after weather and weekday patterns were considered.

The result suggests that mobility around the campus was not explained by weather alone. Teaching, examination and holiday periods added information about why daily volumes changed.

The effect was strongest for pedestrians and weakest for cyclists. Source-level results also varied: Telraam pedestrian and EnCLOD motor-vehicle models improved substantially, while Telraam motor-vehicle results were less consistent.

Pedestrian models100% improved · 2 tests

Cyclist models100% improved · 3 tests

Motor-vehicle models90% improved · 10 tests

Negative binomial models were preferred because daily mobility counts varied more than a Poisson model would allow.