motionlogic - from mobile to mobility





transform communication data
into crowd analytics

Dr Giovanni Marelli @ motionlogic


  1. data types/concepts
  2. people behaviour
  3. tuning/predictions

Data types



Data privacy

  1. handover between cells - calls/internet package
  2. no content/size information
  3. new hashed id every day
  4. no group (age, device, zip...) below 30 people

event collection

event collection
We collect an handover between cells per device (D1-Netz)

Events to movements

event collection
We groups events into activities and trips

Activity probability

event collection
Density probability of activities between cells

User's behaviour

event collection
Different signaling patterns during the day

Daily patterns

event collection
People have different patterns every day

Tourist patterns

event collection
Split nationality, origin, counting activities

age, gender, device types

event collection
Distribution of age classes per zip code

People behaviour

From cells to routes

  1. Interpret cell connection during a trip
  2. Understand mode of transportation
  3. Calculate most probable route
  4. Count people

movements across the country

event collection
How people move during a day

Most frequent origin and destination

event collection
important for local transportation companies


event collection
we work on an efficient routing and infrastructure

local network

event collection
Special modules consider city centers

subway module

event collection
Distinguish the mean of transportation labelling cells (training data)

subway passangers

event collection
Collecting information about commuters




Use of information

  1. activities, footfalls, section counts
  2. points of interest/geo context
  3. statistical geographical population data
  4. public/training reference data

footfall counts

event collection
People movements as feature for commercial success

Intensity of activities

event collection
Activity intensity: breaks of motorway drivers

geographical data

event collection
Validate results with statistical data (population density)

collect important features

event collection
we select the most relevant features

training set

event collection
geographical information to train system


event collection
We train our models to predict success

compare external sources

event collection
Public information from different sources


event collection
We use past time series for forecasting events

mapping on offer

event collection
Calculate the coverage of a commercial activity

whitespot analysis

event collection
Customer potential on regions not covered within a isochrone

Thank you

for your attention


motionlogic website

export pdf