Identifying the five most influential social network accounts for a targeted restaurant marketing campaign in the San Francisco Bay Area.
This project supports a highly targeted marketing campaign for a new restaurant in the San Francisco Bay Area. The core challenge was two-fold: accurately inferring the geographic locations of users with undisclosed information, and mapping the spread of influence through the network to maximize regional reach.
The analysis was conducted using the CRISP-DM framework across a sample of a professional social network.
A scoring algorithm analyzes up to second-degree network connections to predict whether a user is located in the San Francisco Bay Area or elsewhere.
Information spread is modeled similarly to an infection propagation model using a Mean Field simulation to estimate the probability of users seeing and sharing promotional content.
To find the optimal combination of five influencers out of billions of possibilities, a Greedy Algorithm optimized with a Tabu Search metaheuristic is used to avoid local maxima.
| Metric | Result |
|---|---|
| Bay Area users identified | 152 predicted targets |
| Location model accuracy | ~63% (validated against ground-truth) |
| Baseline sharing probability | 10% |
| Campaign exposure rate | 91% across predicted Bay Area audience |
The following five accounts were identified as the optimal combination for maximum Bay Area reach :
['U7024', 'U22747', 'U27287', 'U16141', 'U3955']
Note: Results carry inherent uncertainty due to location prediction accuracy and potential false positives. The 91% exposure rate is based on simulation under a 10% baseline sharing probability.
