In webscale/Telematics: Telematics package Based on AXA data

Telematics Package Examples

This document shows the usage scenarios for the Telematics package. Their are two entry points to using the package. 1) If you do not have a Master Summary File created and you want to create One 2) If you have a Master Summary File and you want to load it and do some analysis

Defining some variable

library(package = "niraj9.telematics")
public.folder = "/home/niraj9/public/"
master.file.3 = paste(public.folder , "master-3.csv" , sep = "")
master.file.4 = paste(public.folder , "master-4.csv" , sep = "")

Scenario 1 - Create a Master Summary File

telematics <- Telematics$new(master.file.4)
arun = telematics$CreateMasterSummary(file.list = "/home/niraj9/public/fileList.csv" ,
                                         root.folder = "/home/niraj9/public/drivers/",
                                         subset.driver.count = 30 ,
                                         cores =4)

Scenario 2 - Load an existing Master Summary File

telematics <- Telematics$new(master.file.3)
telematics$load(verbose = FALSE)

Driver Analysis - Example 1 - Show Drivers loaded

telematics$drivers()

Load a Driver

driver = Driver$new(db = telematics , driver.name = "2751")
driver$summary()

Driver Analysis - Example 2 - Show Driver Graphs

q =driver$ShowCoordinates(trip = 165)
q =driver$ShowVelocity(trip = 165)
q =driver$ShowAcceleration(trip = 165)
q =driver$ShowBreaks(trip = 165)
q=driver$ShowQuantiles(trip = 111)
q=driver$ShowSpeedBreaks(trip = 120)

Driver Analysis - Example 3 - Driver Signature Predictive Model

Create a Unique Driver Signature. i.e a GBM Model that can identify a driver

model = driver$CreateDriverSignature()
summary(model)

Driver Analysis - Example 4 - Driver Class Predictive Model

Use the Telematics class to segment the drivers into 5 driver classes

fit <- telematics$SegmentDrivers(segments = 5)
summary(fit)

Visualize the Driver Clusters

telematics$VisualizeSegments()

webscale/Telematics documentation built on May 4, 2019, 2:05 a.m.