Visualizing New York City WiFi Access with K-Means ClusteringMichael Grogan (MGCodesandStats)BlockedUnblockFollowFollowingFeb 5Visualization has become a key application of data science in the telecommunications industry.
Specifically, telecommunication analysis is highly dependent on the use of geospatial data.
This is because telecommunication networks in themselves are geographically dispersed, and analysis of such dispersions can yield valuable insights regarding network structures, consumer demand and availability.
DataTo illustrate this point, a k-means clustering algorithm is used to analyse geographical data for free public WiFi in New York City.
The dataset is available from NYC Open Data.
Specifically, the k-means clustering algorithm is used to form clusters of WiFi usage based on latitude and longitude data associated with specific providers.
From the dataset itself, the latitude and longitude data is extracted using R:#1.
frame(newyork$LAT,newyork$LON)Here is a snippet of the data:Determine number of clustersNow, the number of clusters need to be determined using a scree plot.
Determine number of clusterswss <- (nrow(newyorkdf)-1)*sum(apply(newyorkdf,2,var))for (i in 2:20) wss[i] <- sum(kmeans(newyorkdf, centers=i)$withinss)plot(1:20, wss, type="b", xlab="Number of Clusters", ylab="Within groups sum of squares")From the above, the curve levels out at approximately 11 clusters.
Therefore, this is the number of clusters that will be used in the k-means model.
K-Means AnalysisThe K-Means analysis itself is conducted:#3.
K-Means Cluster Analysisset.
seed(20)fit <- kmeans(newyorkdf, 11) # 11 cluster solution# get cluster meansaggregate(newyorkdf,by=list(fit$cluster),FUN=mean)# append cluster assignmentnewyorkdf <- data.
cluster <- as.
LAT, color = newyorkdf$fit.
cluster)) + geom_point()In the data frame newyorkdf, the latitude and longitude data, along with the cluster label is displayed:> newyorkdf newyork.
94458 12 40.
94413 13 40.
94517 14 40.
94517 15 40.
94517 16 40.
82075 1181 40.
82105 1182 40.
82106 1183 40.
82175 1184 40.
82178 1185 40.
89455 11Here is a visual illustration:This illustration is useful, but the ideal scenario would be to append these clusters to a map of New York City itself.
Map VisualizationTo generate a map of New York City, the nycmaps library is used, installable from the Github repo as indicated below.
# devtools::install_github("zachcp/nycmaps")library(nycmaps)map(database="nyc")#this should also work with ggplot and ggaltnyc <- map_data("nyc")gg <- ggplot()gg <- gg + geom_map( data=nyc, map=nyc, aes(x=long, y=lat, map_id=region))gg + geom_point(data = newyorkdf, aes(x = newyork.
LON, y = newyork.
LAT), colour = newyorkdf$fit.
cluster, alpha = .
5) + ggtitle("New York Public WiFi")Upon running the above, a map of NYC is generated, along with the associated clusters:This type of clustering gives great insights into the structure of a WiFi network in a city.
For instance, there are 650 separate points in cluster 1, whereas 100 points exist in cluster 6.
This indicates that the geographic region marked by cluster 1 shows heavy WiFi traffic.
On the other hand, a lower number of connections in cluster 6 indicates low WiFi traffic.
K-Means Clustering in itself does not tell us why traffic for a specific cluster is high or low.
For instance, it could be the case that cluster 6 has a high population density, but poor internet speeds result in fewer connections.
However, this clustering algorithm provides a great starting point for further analysis — and makes it easier to gather additional information to determine why traffic density for one geographic cluster might be higher than another.
ConclusionThis example demonstrated how k-means clustering can be used with geographical data in order to visualize WiFi access points across New York City.
In addition, we have also seen how k-means clustering can also indicate high and low density zones for WiFi access, and the potential insights that can be extracted from this regarding population, WiFi speed, among other factors.
Thank you for your time!.