Analyzing Demand Data

1
Analyzing DemandData

• Presented By
• Jonathan D. Washko, BS-EMSA
• Director of Deployment REMSA
• President Washko Associates, LLC
• HPEMS Public Safety Consulting
• Partner Stout Solutions, LLC FirstWatch

2
Analyzing Demand Data

• Discussion Topics
• What is a Demand Analysis
• What kind of data do you need to calculate it
• What are some of the pitfalls to watch out for
• What formulas do you use to calculated it
• What tools do you use to calculate it
• What do you do with this information when
  completed

3
Analyzing Demand Data

• What is a Temporal Demand Analysis?
• A Temporal Demand Analysis (or TDA) is an
  analysis of arrayed and aggregated historical
  call volume by week, hour of day and day of week.
  It is used to help predict and determine the
  number of Quality Unit Hours needed (Demand) for
  each hour of the day and day of week.
• When completed, the analysis will provide
  staffing needs for a total of 168 hours (total
  number of hours in a week). From this analysis,
  a Peak Load Staffing Schedule can be built to
  match the prediction model (Matching Supply with
  Demand).

4
Analyzing Demand Data

• Temporal Demand Analysis Fundamental Assumptions
• Assumes Each Call Takes one hour to complete (11
  S/D Ratio)
• Needs to be adjusted to each system accordingly
• Use Task Time to adjust as needed if average is
  gtlt 60 minutes
• Systems with lower Task Times require less
  resources
• Systems with higher Task Times require more
  resources
• Adjustments can be made through demand
  multipliers or the performing of a Task Time TDA
  (A much more complex analysis)

Efficiency Alert! Controlling your systems Task
Time can have a HUGE financial impact on your
system staffing costs so long as controls are
kept to balance the triad.
Pitfall Alert! Inaccurate Task Time calculations
can substantially impact the outcome of a demand
analysis and put patient lives or an organization
at risk. Perform the Task Time Analysis with due
diligence and caution ensuring accuracy and
validity!
5
Analyzing Demand Data

• Data Set Characteristics
• Bad in / bad out concept
• What to measure and why
• Requests, Responses or Transports?
• Call Priorities to include or exclude
• Standby / Special Events
• Multi-Unit Responses
• Other Variables (CCT, Specialized Units, Special
  Calls, Special Circumstances, etc.)

6
Analyzing Demand Data

• Other Things You Need to Know
• Desired response time reliability percentage
• Inefficiency (LUH) buffer / cushion
• Call volume seasonality
• Some Art (SWAG)
• Response time requirements
• Response time zone balancing requirements
• Effects of city infrastructure (or lack there of)
• Effects of traffic patterns
• Effects of political Posts
• Effects of other unique system anomalies

7
A Temporal Demand Analysis for Monday
8
A Temporal Demand Analysis for Monday
Raw Demand Analysis Data. P1, P2, P3, P4 P7
Count of responses that arrived on scene by hour
of day, day of week, chronologically ordered by
date. A total of 20 weeks worth of most recent
data from the CAD system.
9
A Temporal Demand Analysis for Monday
Military Date Format of Arrayed Days (Mondays)
in Chronological Order In this case the date is
Monday February 03, 2003
Raw Demand Analysis Data. P1, P2, P3, P4 P7
Count of responses that arrived on scene by hour
of day, day of week, chronologically ordered by
date. A total of 20 weeks worth of most recent
data from the CAD system.
10
A Temporal Demand Analysis for Monday
Hours of Day in Hour Ending Format e.g. 21
2000 through 2100
Raw Demand Analysis Data. P1, P2, P3, P4 P7
Count of responses that arrived on scene by hour
of day, day of week, chronologically ordered by
date. A total of 20 weeks worth of most recent
data from the CAD system.
11
A Temporal Demand Analysis for Monday
Total of All Hours for Each Week (Totaled
Across) In this case, there were 196 Responses on
Feb. 10, 2003
Raw Demand Analysis Data. P1, P2, P3, P4 P7
Count of responses that arrived on scene by hour
of day, day of week, chronologically ordered by
date. A total of 20 weeks worth of most recent
data from the CAD system.
12
A Temporal Demand Analysis for Monday
Represents that on February 17, 2003 there were
13 Responses between 1100 and 1200
Raw Demand Analysis Data. P1, P2, P3, P4 P7
Count of responses that arrived on scene by hour
of day, day of week, chronologically ordered by
date. A total of 20 weeks worth of most recent
data from the CAD system.
13
A Temporal Demand Analysis for Monday
A Temporal Demand Analysis for Mondays
Demand Analysis Analytics. Used to calculate
the required number of Quality Unit Hours
(Demand) by Hour of day for this particular day
of the week (In this case, Monday) There are
various statistical methods used to calculate
system demand, all are accurate and correct.
Experience has shown that Average Peak (a formula
created by Jack Stouts team) consistently yields
an accurate prediction of the 90th Percentile of
demand.
Section II EMS Production Model Art Science,
IV.c
14
A Temporal Demand Analysis for Monday
Total Responses for the Hour Totaled by column
(hour) In this case the total is 72 responses for
0300 to 0400 XL Formula sum(CRCR)
15
A Temporal Demand Analysis for Monday
Totals are for entire 20 week period (All Hours
all Mondays) XL Formula sum(CRCR)
16
A Temporal Demand Analysis for Monday
Min or Minimum number of Calls for the 20 week
period for the hour of day and day of week. In
this case the Min for 0300 to 0400 1 (Which
occurs twice during the 20 week period) XL
Formula min(CRCR)
17
A Temporal Demand Analysis for Monday
Total of Min or Minimum number of Calls by hour
for all hours. Totaled by row (horizontally) In
this case the Total Min 79 XL Formula
sum(CRCR)
18
A Temporal Demand Analysis for Monday
Max or Maximum number of Calls for the 20 week
period for the hour of day and day of week. In
this case the Max for 0300 to 400 9 (Which
occurs once during the 20 week period) XL
Formula max(CRCR)
19
A Temporal Demand Analysis for Monday
Total of Max or Maximum number of Calls by hour
for all hours. Totaled by row (horizontally) In
this case the Total Max 357 XL Formula
sum(CRCR)
20
A Temporal Demand Analysis for Monday
Average number of Calls for the 20 week period
for the hour of day and day of week. In this case
the Average for 0300 to 400 3.6 XL Formula
average(CRCR)
21
A Temporal Demand Analysis for Monday
Total of Average number of Calls by hour for all
hours. Totaled by row (horizontally) In this
case the Total Average 198.8 XL Formula
sum(CRCR)
22
A Temporal Demand Analysis for Monday
TMT (Total Mission Time) Multiplier is used to
adjust demand calculations by taking the number
and multiplying it times the Demand Measurement.
Example if your Average TMT 75 Minutes then
Take 75 / 60 (minutes in an hour) 1.25 This
means that your Demand Measurement will be
adjusted by 1.25 or 125 In this case the TMT
Multiplier 0 for simplicity
23
A Temporal Demand Analysis for Monday
Average of TMT Multiplier for all hours of this
day. Totaled by row (horizontally) In this case
the Average 0.0 XL Formula Average(CRCR)
24
A Temporal Demand Analysis for Monday
The Standard Deviation of Responses for the hour.
In this case the Standard Deviation for 0300
to 0400 2.0 XL Formula stdev(CRCR)
25
A Temporal Demand Analysis for Monday
Total of Standard Deviation for all hours of this
day. Totaled by row (horizontally) In this case
the total Standard Deviation 72.6 XL Formula
Sum(CRCR)
26
A Temporal Demand Analysis for Monday
The Average High is a Stoutian Measurement that
represents approximately the 75th percentile of
demand. It is calculated by taking the maximum
number of calls in each consecutive 5 4 week
periods of a 20 week analysis then dividing the
sum of these number by 5 (or average of the 5
periods) In this example, the Average High for
0300 to 0400 5.8 The XL Formula
(Max(CRCR) Max(CRCR) Max(CRCR)
Max(CRCR) Max (CRCR)) / 5 The resultant is
then multiplied by the TMT Multiplier for TMT
Adjustments
27
A Temporal Demand Analysis for Monday
Total of Average High for all hours of this day.
Totaled by row (horizontally) In this case the
total Average High 277.4 XL Formula
Sum(CRCR)
28
A Temporal Demand Analysis for Monday
The 90th Percentile (or X Percentile) is a
statistical ranking method used to determine the
demand at X percent. In this case the 90th
Percentile Rank for 0300 to 0400 6.1 XL
Formula percentile(CRCR,.9) where.9 90 The
resultant is then multiplied by the TMT
Multiplier for TMT Adjustments
29
A Temporal Demand Analysis for Monday
Total of 90th Percentile for all hours of this
day. Totaled by row (horizontally) In this case
the total 90th Percentile 280.1 XL Formula
Sum(CRCR)
30
A Temporal Demand Analysis for Monday
The Average Peak is a Stoutian Measurement that
represents approximately the 90th percentile of
demand. It is calculated by taking the maximum
number of calls in each consecutive 2 10 week
periods of a 20 week analysis then dividing the
sum of these number by 2 (or average of the 2
periods) In this example, the Average Peak for
0300 to 0400 8.0 The XL Formula
(Max(CRCR) Max(CRCR) ) / 2 The resultant is
then multiplied by the TMT Multiplier for TMT
Adjustments
31
A Temporal Demand Analysis for Monday
Total of Average Peak for all hours of this day.
Totaled by row (horizontally) In this case the
total Average Peak 321.5 XL Formula
Sum(CRCR)
32
A Temporal Demand Analysis for Monday
The Smoothed Average Peak is a statistical
smoothing of the Average Peak and is used to
blend the severity of hour to hour demand
fluctuations for easier peak-load scheduling. It
is calculated by taking 20 of the previous hour
(Blue) 60 of the current hour (Red) 20 of
the next hour (Yellow). In this example,
Smoothed Average Peak for 0300 to 0400
7.6 The XL Formula (CR.2)(CR.6)(CR.2)
33
A Temporal Demand Analysis for Monday
Total of Smoothed Average Peak for all hours of
this day. Totaled by row (horizontally) In this
case the total Smoothed Average Peak 319.7 XL
Formula Sum(CRCR)
34
A Temporal Demand Analysis for Monday
2x StdDev Mean is a Statistical Process Control
methodology used by industry as an upper control
chart limit. It represents approximately the 95
of demand. It is calculated by taking the
Standard Deviation (Red) calculated previously
for the hour, multiplying it by 2 then adding it
to the previously calculated average (Blue). In
this example, 2x StdDev Mean for 0300 to
0400 7.67 The XL Formula (CR2)(CR) The
resultant is then multiplied by the TMT
Multiplier for TMT Adjustments
35
A Temporal Demand Analysis for Monday
Total of 2x StdDev Mean for all hours of this
day. Totaled by row (horizontally) In this case
the total 2x StdDev Mean 344.0 XL Formula
Sum(CRCR)
36
A Temporal Demand Analysis for Monday
The UH Adj / Eff Buffer is used to allow managers
the ability to adjust up or down their staff to
lines based on individual system needs. This is
where some of the art comes in helping to
determine this figure. This can also be used to
adjust the demand for contractual unit minimums
and other needs. In this case, the UH Adj / Eff
Buffer for 0300 to 0400 2
37
A Temporal Demand Analysis for Monday
Total of UH Adj / Eff Buffer for all hours of
this day. Totaled by row (horizontally) In this
case the total UH Adj / Eff Buffer 48.0 XL
Formula Sum(CRCR)
38
A Temporal Demand Analysis for Monday

• Adjusted Supply / Inefficiency Buffer

• UH Adjustment Variables
• The part of the Art of HPEMS (SWAG)
• Response Time Requirements
• Infrastructure Impacts
• Geography / Coverage Area
• System Idiosyncrasies and Their Impacts
• Lost Unit Hours
• Politics
• Lots of Others

39
Demand Curve Adjustments
Example Average Peak 4 Minimal Staffing
40
A Temporal Demand Analysis for Monday
The Staff To line is the number of QUH required
for that hour of that day. It typically defines
the minimum number of expected demand at X
Percentile (Based on demand formula used). In
this example, we are using the Average Peak UH
Adj / Eff Buffer as our Staff to Calculation. The
definition of this calculation varies by system
and is not set. Frequently HPEMS systems use the
Smoothed Average Peak UH Adj / Eff Buffer
approach. In this Example the Staff to Line for
0300 to 0400 10.0
41
A Temporal Demand Analysis for Monday
Total of Staff to Line for all hours of this day.
Totaled by row (horizontally) In this case the
Staff To Line 369.5 XL Formula Sum(CRCR)
42
Analyzing Demand Data

• So What Do All These Numbers Mean?

43
Analyzing Demand Data

• So What Do All These Numbers Mean?

• The Temporal Demand Analysis results in telling
  you how many Quality Unit Hours need to be
  produced by hour of day and day of week.
• Each hour of each day will have its own number
  for a total of 168 Production / Demand
  Requirements
• You build your Production Schedule to match
  these numbers

44
Analyzing Demand Data

• So What Do All These Numbers Mean?

• The total columns come in handy as by adding the
  appropriate total for each day will give you the
  total QUH requirement for the week.
• This number can then be used to calculate FTEs,
  UHU and other budget figures.
• Keep in mind this is the optimal requirement. It
  is very difficult (if not near impossible) to
  create a schedule that exactly matches the demand
  requirements.
• The best Schedule to Demand matching I have
  experienced is /- 2 to 3 variation.
• Zolls Resource Planner allows you to take Demand
  Data from this analysis to easily build a
  schedule that ACCURATELY matches your systems
  Demand (see next session).

45
Analyzing Demand Data

• Tricks of the Trade

• If your system is responsible for multiple
  geographic regions that are not contiguous or
  operate as separate deployment centers, a
  separate TDA for each area may be warranted
• Microsoft Excel or Microsoft Access (for advanced
  programmers) is the best tool for creating a TDA
• Microsoft Excel Pivot Tables allow for immediate
  arraying of data into tables that can then be
  copied and pasted into a TDA template

46
Analyzing Demand Data

• Tricks of the Trade

• You really only need a few key pieces of data to
  calculate the TDA raw data array
• TDA
• Priority of the call
• Call clock start date and time
• Hour of day and day of week calculations can be
  performed from this information
• GDA (plusTDA data)
• Longitude of the call (or full address if not
  available)
• Latitude of the call (or full address if not
  available)

47
Analyzing Demand Data

• Tools That Can Help

• FirstWatch
• FirstWatch (Data Surveillance Tool) has created a
  Demand Analysis Tool that allows for a button
  push assessment of demand that has an output
  file that can be directly imported into Zolls
  Resource Planner www.first-watch.us based on the
  templates used in this session.
• Zoll Data Systems / Washko Associates
• Working on developing a Resource Planner Add On
  that will allow you to import demand data
  directly into Resorce Planner
• Developed an exclusive arrangement for customers
  looking for additional expertise and help (Speak
  with me or your Zoll Sales Representitive)

48
Analyzing Demand Data

• Questions / Discussion