Origin and penetration algorithms

We determine where our population lives or works by identifying behavioural patterns.

A person will be identified as living in a particular neighbourhood when he or she has been seen several times and on most days of the month in the evening and morning hours in the same neighbourhood. 

He or she will be identified as working in a particular neighbourhood if he or she has been seen repeatedly during the hours of a standard workday (10 a.m. to 6 p.m. in France) and on weekdays.

For origin, we then analyse the percentage of people living in each neighbourhood out of the total population seen in or near the object analysed and whose home or workplace we know. The origin indicator takes into account the recurrence of unique daily visits during the week.

For example: a person who has visited this address every day of the month will weigh 30 times more in our statistics than a person who has visited it only once.

For the penetration rate indicator, we compare the number of known inhabitants of a given neighbourhood who have visited the asset to the total number of known inhabitants of that neighbourhood.

Here, the recurrence of visits is not taken into account to have rates that never exceed 100% and give an idea of the percentage of the population that has visited the asset under analysis at least once.

For reasons of reliability and confidentiality, we also do not display neighbourhoods with fewer than 20 inhabitants in our databases in a given month. This restriction applies to certain cities with small populations (France: < 1,400 inhabitants, Italy: < 2,100 inhabitants, Spain: < 1,100 inhabitants. No restrictions for other countries).

A series of adjustments is also applied so that our statistics are as close as possible to the reality on the ground (geographical adjustment, socio-demographic identification bias, etc.

Why does the analysis show 80% of the cumulative origin of visitors?

Residual origins: Beyond the 80% threshold, the rates become insignificant. Our margins of error no longer allow us to ensure the reliability of the rates displayed. At Mytraffic, we want to ensure the reliability of our data above all. We make sure that our rates are calculated on a minimum number of GPS points to be statistically representative.

Minimal rates: Beyond the 80% threshold, the rates become minimal and very insignificant. They are not suitable for displaying or understanding the data. Ex: 0.00001% of origin

Scattered origins: Beyond the 80% threshold, the rates become very scattered over the territory and therefore difficult to analyse or to mobilise in the context of using the platform as an assistance in decision-making (for example, on the Gare de Lyon polygon, the 100% origin becomes so scattered that it covers practically the whole of France).

Exceptional and inconstant origins: Above the 80% threshold, the majority of origins are linked to exceptional visits. The majority of neighbourhoods that appear in a given month are not part of the origin area of the same asset in other months. This makes it impossible to carry out a follow-up analysis and monitor the evolution of the rates:. e.g. Presence in the polygon of a person who has come to visit his family from the other side of the country.

Page loading time: This 80% threshold also allows us to optimise the loading time of our platform for better navigation.