Direction Direction = 0: Northbound Direction = 1: Southbound.

Slides:



Advertisements
Similar presentations
 Will help you gain knowledge in: ◦ Improving performance characteristics ◦ Reducing costs ◦ Understand regression analysis ◦ Understand relationships.
Advertisements

Matt’s Schedule. Headway Variation Estimated Load vs. Passenger Movement.
1 Reliability-Based Timepoint Schedules for Long Headway Transit Routes Peter G. Furth, Northeastern University with Theo H.J. Muller, Delft University.
Time-Series Analysis and Forecasting – Part III
1 Multiple Regression Response, Y (numerical) Explanatory variables, X 1, X 2, …X k (numerical) New explanatory variables can be created from existing.
Seasonal Models Materials for this lecture Lecture 3 Seasonal Analysis.XLS Read Chapter 15 pages 8-18 Read Chapter 16 Section 14.
Warm Up 1)Make two graphs that demonstrate your knowledge of the vertical line test – one of a function and one that is not a function. 2)The length of.
 Often times meteorologists and climatologists to make some calculations in order to better understand trends in the weather. Here are three calculations.
Short-Term Load Forecasting: Similar Day-Based Wavelet Neural Networks Ying Chen, Peter B. Luh, Fellow, IEEE, Che Guan, Yige Zhao Laurent D. Michel Matthew.
Seasonal Models Materials for this lecture Lecture 3 Seasonal Analysis.XLS Read Chapter 15 pages 8-18 Read Chapter 16 Section 14.
ForecastingModelsWith Trend and Seasonal Effects.
Section 10-3 Chapter 10 Correlation and Regression Correlation
Chapter 37 Presentation of Data 2. Learning Objectives Read from a time-series graph Read from a time-series graph Draw a time-series graph Draw a time-series.
Notes on Data Collection and Analysis Dale Weber PLTW EDD Fall 2009.
Geostatistical approach to Estimating Rainfall over Mauritius Mphil/PhD Student: Mr.Dhurmea K. Ram Supervisors: Prof. SDDV Rughooputh Dr. R Boojhawon Estimating.
1 1 Slide © 2012 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole.
Seasonal Models Materials for this lecture Lecture 9 Seasonal Analysis.XLSX Read Chapter 15 pages 8-18 Read Chapter 16 Section 14 NOTE: The completed Excel.
ERCOT Long-Term Demand and Energy Forecasting February 20, 2007 Bill Bojorquez.
Section 12.2 Linear Regression HAWKES LEARNING SYSTEMS math courseware specialists Copyright © 2008 by Hawkes Learning Systems/Quant Systems, Inc. All.
EXAMPLE 3 Use sinusoidal regression Energy The table below shows the number of kilowatt hours K (in thousands) used each month for a given year by a hangar.
Time series Decomposition Farideh Dehkordi-Vakil.
INTERNAL ACHIEVEMENT STANDARD 3 CREDITS Time Series 1.
Number of Movies Frequency TOTAL: 88 The following data shows the number of movies 88 students watched one week during the summer.
Stopping distance Subtitle. Some Review questions and topics You should know the content of these slides. Computations can be done on your calculator.
Exploring Seasonal Variance among Light Sweet Crude and Corn Futures Caleb Seeley 2/27/08.
REGRESSION DIAGNOSTICS Fall 2013 Dec 12/13. WHY REGRESSION DIAGNOSTICS? The validity of a regression model is based on a set of assumptions. Violation.
Examining Bivariate Data Unit 3 – Statistics. Some Vocabulary Response aka Dependent Variable –Measures an outcome of a study Explanatory aka Independent.
Comprehensive Operations Analysis LYNX 2013 Comprehensive Operational Analysis John Lewis, CEO.
Deseasonalizing Forecasts
Chapter 3-Examining Relationships Scatterplots and Correlation Least-squares Regression.
Guest Lecture weather briefing -- climate -- compositing -- paper review -- guest of the guest lecture -- wrap up.
Unit 4 Lesson 5 Demonstrating Mastery M.8.SP.1 To demonstrate mastery of the objectives in this lesson you must be able to:  Construct and interpret.
Bus Service Reliability Developing new performance measures Unreliability Vs. Bus Bunching Meead Saberi K.
Stat 31, Section 1, Last Time Distributions (how are data “spread out”?) Visual Display: Histograms Binwidth is critical Bivariate display: scatterplot.
Frap Time! Linear Regression. You will have a question to try to answer and I’ll give you 15 minutes. Then we will stop, look at some commentary then.
09/17/2006 Ken Donohoo ERCOT Peak Day August Initial Settlement Data by Fuel Type.
Transit Signal Priority: The Importance of AVL Data David T. Crout Tri-County Metropolitan Transportation District of Oregon (TriMet) Presented at Transportation.
Complete Graph. What is wrong with the Graph? The graph needs to have numeric labels on the axes. We can not determine a coordinate without them. Does.
Chapter 14 STA 200 Summer I Scatter Plots A scatter plot is a graph that shows the relationship between two quantitative variables measured on the.
Coordinate Algebra Practice EOCT Answers Unit 4. #1 Unit 4 This table shows the average low temperature, in ºF, recorded in Macon, GA, and Charlotte,
Lecture 9 Seasonal Models Materials for this lecture Lecture 9 Seasonal Analysis.XLSX Read Chapter 15 pages 8-18 Read Chapter 16 Section 14.
TIME SERIES ‘Time series’ data is a bivariate data, where the independent variable is time. We use scatterplot to display the relationship between the.
 This lesson covers two methods for finding an equation for a line that roughly models a set of data.  The first way is to eyeball a possible line,
2016 Fire Season Considerations and Outlook SWCC Predictive Services Updated February 17, 2016.
Multi-Area Load Forecasting for System with Large Geographical Area S. Fan, K. Methaprayoon, W. J. Lee Industrial and Commercial Power Systems Technical.
Eduardo Acevedo IME 301 Winter 2010 Time on the Road.
Correlation of a Scatter Plot. What is correlation… O It describes the direction and strength of a straight-line relationship between two quantitative.
Section 12.2 Linear Regression
Critical Analysis.
Calculating Median and Quartiles
AP Stats: 3.3 Least-Squares Regression Line
Chapter 3: Describing Relationships
Forecasting Models With Trend and Seasonal Effects
Chapter 3: Describing Relationships
Chapter 3: Describing Relationships
Coordinate Algebra Practice EOCT Answers Unit 4.
Chapter 3: Describing Relationships
3.1 ONE AND TWO VARIABLE DATA
Chapter 3: Describing Relationships
Homework: pg. 180 #6, 7 6.) A. B. The scatterplot shows a negative, linear, fairly weak relationship. C. long-lived territorial species.
Chapter 3: Describing Relationships
Chapter 3: Describing Relationships
Scatterplots, Association and Correlation
Coordinate Algebra Practice EOCT Answers Unit 4.
Coordinate Algebra Practice EOCT Answers Unit 4.
Chapter 3: Describing Relationships
Chapter 3: Describing Relationships
Today we are going back in time to help us predict the future!!!
Utah’s Climate CLIMATE IS: weather patterns over a long period of time
Homework: PG. 204 #30, 31 pg. 212 #35,36 30.) a. Reading scores are predicted to increase by for each one-point increase in IQ. For x=90: 45.98;
Presentation transcript:

Direction Direction = 0: Northbound Direction = 1: Southbound

Trip Time Points to include in the paper Trip time is a very powerful tool. Important to generate regularly and for every route. Can use to understand the effect of different variables on trip time. Can give the example of considering the impact of adding another stop, etc.

Headway Variation

Length of Dwell

Early/Late

Weather

Interesting to note the below average passenger boardings in the summer and x-mas week Need to calculate the average by quarter or by month, since the summer is a distinct season

Dwell Time Scatter Plots

Dwell 3-D

Dwell 3-D Axes Reversed

Dwell Histogram (dwell > 0) outliers not removed from mean

Dwell Regression Dwell <= 1 min, Boardings Only X1 = Boardings X2 = Alightings X3 = Late (> 3 minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp N = 737,614

Dwell Regression Dwell <= 1 min, Boardings Only, Late/Early <= 15 minutes X1 = Boardings X2 = Alightings X3 = Late (in minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp N = 733,005

Dwell Regression Dwell <= 1 min, Boardings Only X1 = Boardings X2 = Alightings X3 = Late (> 3 minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp X7 = Boardings 2 X8 = Alightings 2 N = 737,614

Dwell Regression Dwell <= 1 min, Boardings Only, Late/Early <= 15 minutes X1 = Boardings X2 = Alightings X3 = Late (in minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp X7 = Boardings 2 X8 = Alightings 2 N = 733,005

Dwell Regression Dwell <= 1 min, Alightings Only X1 = Boardings X2 = Alightings X3 = Late (> 3 minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp N = 876,802

Dwell Regression Dwell <= 1 min, Alightings Only, Late/Early <= 15 minutes X1 = Boardings X2 = Alightings X3 = Late (in minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp N = 864,491

Dwell Regression Dwell <= 1 min, Alightings Only X1 = Boardings X2 = Alightings X3 = Late (> 3 minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp X7 = Boardings 2 X8 = Alightings 2 N = 876,802

Dwell Regression Dwell <= 1 min, Alightings Only, Late/Early <= 15 minutes X1 = Boardings X2 = Alightings X3 = Late (in minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp X7 = Boardings 2 X8 = Alightings 2 N = 864,491

Dwell Regression Dwell <= 1 min, Both Boardings & Alightings X1 = Boardings X2 = Alightings X3 = Late (> 3 minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp N = 634,402

Dwell Regression Dwell <= 1 min, Both Boardings & Alightings, Late/Early <= 15 minutes X1 = Boardings X2 = Alightings X3 = Late (in minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp N = 627,022

Dwell Regression Dwell <= 1 min, Both Boardings & Alightings X1 = Boardings X2 = Alightings X3 = Late (> 3 minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp X7 = Boardings 2 X8 = Alightings 2 N = 634,402

Dwell Regression Dwell <= 1 min, Both Boardings & Alightings, Late/Early <= 15 minutes X1 = Boardings X2 = Alightings X3 = Late (in minutes) X4 = Timepoint (dummy) X5 = Precipitation X6 = Ave Temp X7 = Boardings 2 X8 = Alightings 2 N = 627,022

Trip Time Model Modified Ahmed Version – outliers removed tripmiles > 0 & tripmiles 0 X1 = Distance (in miles) X2 = Scheduled Number of Stops X3 = Direction or Southbound X4 = AM Peak X5 = PM Peak X6 = Actual Number of Stops X7 = Boardings + Alightings X8 = Lift X9 = Average Passenger Load X10 = Total Dwell Time X11 = Precipitation X12 = Average Temperature X13 = Summer (dummy variable if month = June thru August) X14 = (Boardings + Alightings) 2 Ahmed says use this version N = 53,072 (# of trips)

Trip Time Model Modified Ahmed Version – outliers removed tripmiles > 0 & tripmiles 0 X1 = Distance (in miles) X2 = Scheduled Number of Stops X3 = Direction or Southbound X4 = AM Peak X5 = PM Peak X6 = Actual Number of Stops X7 = Boardings + Alightings X8 = Lift X9 = Average Passenger Load X10 = Total Dwell Time X11 = Precipitation X12 = Average Temperature X13 = Summer (dummy variable if month = June thru August) X14 = (Boardings + Alightings) 2 This is the same as the prior model except this one is not excluding runs where difference between boardings and alightings is greater than 100 N = 53,130 (# of trips)

Tips for Utilizing a Large Data Set Managing a data set in excess of 6 million rows was a challenge that required careful thought and experimentation. Ultimately, we were able to arrange the data such that complex calculations could be performed very quickly. Even with the added capacity of Excel 2007, 6 million rows of data cannot be opened. Instead, we imported the data into a statistical software application called Matlab™. Initial calculations, however, were prohibitively slow. For example, with more than 54,000 unique bus trips, looping through each trip in order to calculate trip level information such as trip distance, trip time, etc. took several hours. The keys to our ultimately success were 1) sorting the data and 2) setting up variables and scripts that optimized Matlab’s powers of calculation. Below is are examples of the basic logic we used. The first step was to create a unique identifier for each trip, which we called ‘bus_run’. The combination of fields in the data that make up a unique trip are described earlier in the paper. The ‘unique’ function in Matlab not only identifies the unique values in a variable, it is also capable of indexing the first and last points in the data where each value occurs. In the script below, the indexed location is stored in the ‘low’ and ‘hi’ variables respectively. These two lines of code calculate in a matter of seconds. % create a unique trips variable [uniqueTrips,low,n] = unique(bus_run, 'first'); [uniqueTrips,hi,n] = unique(bus_run, 'last'); Now we can calculate the total dwell time (and several other variables) for each trip by summing up all of the dwells in each trip. The ordering of the data by trip and the fact that we know the location of the first and last occurance of each unique trip (‘low’ and ‘hi’ variables) enables us to loop through 54,000 trips and 6 million records in a matter of seconds. for x = 1:length(uniqueTrips) ave_load_perrun(x) = mean(estimated_load(low(x):hi(x))) ; total_dwell(x) = sum(dwell(low(x):hi(x))); total_ons(x) = sum(ons(low(x):hi(x))); total_offs(x) = sum(offs(low(x):hi(x))); end Others calculations do not require a for loop and can be made almost instantaneously. For example, the ‘starttime’ of each trip is equal to the ‘leave_time’ of the first occurance of each unique trip. As written below, this variable contains 54,311 records, each one indicating the time that the respective trip began its service. x = 1:length(uniqueTrips); starttime(x) = leave_time(low(x)); Similarly, the total trip time for each trip was calculated by subtracting the ‘leave_time’ on the last instance of a trip from the ‘leave_time’ of the first instance. Again, the result is a trip level variable with 54,311 records. Total trip mileage is calculated in a similar way. % total trip time per run triptime(x) = leave_time(hi(x)) - leave_time(low(x)); % total trip miles per run tripmiles(x) = train_mileage(hi(x)) - train_mileage(low(x));

Trip Time Model Am Peak, Direction = 0

N = NSTT = x(sec)