Data release note for v5.0

V5.0 has been released!

Release date: 09 November 2026

We're constantly working to improve our estimates. In addition to updating our statistics on a monthly basis, we will also be releasing new versions periodically to bring innovation and corrections to our methodology, and provide you with the best datasets possible. 

We’re pleased to launch v5.0 estimates. Here's a summary of the changes:

Key take-aways:

• V5.0 updates all estimates for monthly residents and relocations (data from January 2020 to now) due to the use of new input data 
• June 2025 estimates, previously missing, are now included in the times series. 
• There is a minor change to the methodology to estimate residents and relocations (see below)
• We’ve also revised some of the terminology used on the platform 

Updates

• New input data
• v5.0 uses CDR aggregates of all event types: calls, SMS, and now including mobile data session (MDS) events. With more event types included in the aggregates, we see an increase in the number of detectable Digicel users and therefore improved coverage of the population when compared to using only calls and SMS events (v4.0 and prior).
• Calculation of weights: 
• Methodologically, we’ve made a minor change to the calculation of weights to help better attenuate for large changes and deal with longer data gaps (see Methodology Report v5.0 for more details). 
• Revised terminology 
• Total inflows (“Total incoming” previously)
• Total outflows (“Total outgoing” previously)
• Total net flows (“Incoming minus outgoing” previously)
• Travellers (“Movements” previously)
• Anomaly (“Abnormality” previously)
• Experimental (occasionally “Demo/experimental” or “trial” previously)
• Estimates (occasionally “Indicators” or “metrics” previously)

Differences between residents and relocations estimates v4.0 and v5.0

V5.0 estimates overwrite and replace v4.0 entirely, including each communal section’s baseline population estimate (though the country’s overall baseline population estimate is constant between v4.0 and v5.0). The resulting changes to the estimates are relatively small for a majority of communal sections. The most notable changes are found in communal sections where the baseline estimate was revised upwards in the case that the estimated number of active Digicel users far exceeded the existing population estimate in January 2020. With over six years of monthly estimates in the time series now, other large differences between the previous and latest versions of the estimates are often the result of differences in long- to medium-term trends that  began at an inflection point earlier in the time series, with the gap between the two versions of estimates gradually widening over time.

If you have any questions about these changes and our estimates, please have a look at our methodology here, or contact us at haiti.mobility-dashboard@flowminder.org.