Project 1J: Time of Day of 911 Calls

Overview

This data set contains times and dates of all Seattle Fire Department 911 dispatches in 2015.

Details

The data were processed from the data available at: https://data.seattle.gov/Public-Safety/Seattle-Real-Time-Fire-911-Calls/kzjm-xkqj which provides Seattle Fire Department 911 dispatches and is updated every 5 minutes. The calls with a time stamp from 2015 are included in this data.

Data Description

Variable	Description
Type	type of aid request (such as “Aid Response” and “Medic Response”)
Datetime	Timestamp of the call
Latitude	Latitude of the call’s source
Longitude	Longitude of the call’s source
date	Date extracted from Timestamp
date_in_days	Date converted to an integer from 1 to 365, for convenience
time_of_day_as_fraction	Time of day converted to a number from 0 to 1

seattle_911 = read.csv("data/seattle_groupJ_time_of_calls.csv")
head(seattle_911)

Data Files

• seattle_groupJ_time_of_calls.csv

Objectives

We will consider the distribution of the times of day when the calls arrive. Are 911 calls uniform across the day and night or do they exhibit some pattern? The goal is to find a distribution that fits the times of day of calls. Note: the variable time_of_day_as_fraction is there for ease – you can compare it to the time in the variable Datetime. The variable time_of_day_as_fraction should be near 0 in the very early morning, near 0.5 at noon, and near 1 in the evening.

• Before focusing on times of day of the calls, first plot the Longitude (x-axis) against the Latitude (y-axis) for the calls’ sources. Can you explain the patterns you observe in this plot?
• Now turn your attention to the time of day of the calls. Provide numerical and graphical summaries of the time of day of the calls.
• Consider fitting at least 1 known distribution discussed in class (this semester and last semester) to this variable. For each distribution you consider, explain how you are estimating the relevant parameters (e.g., are you using the MLEs? MOMs? etc.).
• Try shifting the data by 12 hours (i.e., so that the time range goes from noon to noon rather than midnight to midnight) and consider fitting at least 1 known distribution to the shifted data.
• Provide some assessment of fit of the distributions to the data. For example, q-q plots are useful for this purpose.
• Make some conclusions about modeling this variable.