Tomorrows bookings for Tourism
Project Profile
The client wanted to explore the possibility to train a predictive model based on movement data, which can estimate a visitor density based on the origin market.
The POIS were defined by the client and coordinated with the mobile operator for signaling capabilities.
The experience in handling movement data as well as the technical estimability of its use by the customer was to be evaluated.
A small POC ecosystem between motion data and PowerBI visualization was created.
At a glance - essential project data
| Duration | From 4/1/2020 to 7/30/2020 with about 4 months of full engagement | |
|---|---|---|
| Data and Tools | Market - Tourism Sources • Mobile phone data • Flight data VIE • Weather • Global vacation calendar • Power-BI | |
| Integration | • Web API for Model Usage • CSV on FTP • PowerBI Datasets | |
| AI Methods • Deep - Learning • Time Series |
Engagement Use-Case
Predict tourist visit history at locations or POIs over the course of the day over the booking year.
Client motivation / Solution aims
Increase efficiency of occupancy rates at a POI
Touristic forecasting of future visitation trends in tourist markets
Dynamic pricing of events
Occupancy forecasting and budgeting for POIs
Campaign optimization for visitor control
AI Approach
| AI key technology used in our solution | Adjustment of mobile signaling due to structural breaks (roaming contracts, outages, new masts, etc). Adjusting the average length of stay per POI. Qualification of the utilization by Google "popular times" | |
|---|---|---|
| Solution Approach | • Multiple DeepLearning Models • DNN • TimeSeries Prediction | |
| Project Approach | Simply agile | |
| Project Type | Proof-Of-Concept (POC) | |
| ML Integration and ML Operations | • Operation Integration API • Visualization Power-BI |
Insights and Details
Dwell time at POI
Feature Importance for optimization of model parameters
Prediction of SIM population by country of origin and time of creation
