ATR-TDA2, SC-TDA2: User Manual

Transcription

1 Minnesota Department of Transportation Office of Transportation Data Forecast and Analysis ATR-TDA2, SC-TDA2: User Manual Mar 25, 2014, Original Publication Dec, 2017, revised June, 2018, revised By Dr. Take M. Kwon University of Minnesota, Duluth

2 Contents 1. Introduction ATR-TDA Brief Implementation Description How to Use the ATR-TDA2 Software Plot Utility of ATR-TDA ATR Data and Log File Format SC-TDA Brief Implementation Description How To Use the SC-TDA2 Program Output File Data Formats Data Plot Tab Concluding Remark

3 1. Introduction This manual describes two software packages, ATR-TDA2 and SC-TDA2, developed by Dr. Taek Kwon at the University of Minnesota Duluth (UMD) for MnDOT. These software packages are available in MSI formats, which are installable in PC by a simple double click. For software developments, a version of Microsoft Visual Studio with.net Framework 4.5 was used. One of the critical aspects of using the ATR and SC software packages is the assumed file structure used by all UMD developed software packages. A specific directory tree with specific directory names must be manually created before using the software. The directory tree and names should exactly follow Figure 1 without change of the letter case neither the spelling. First, the root directory named traffic must be created. Inside the root directory, two directories should be created, which are defines and processed. Inside the defines or processed directory, ATR and SC directories should be available. The rest of subdirectories and files are automatically created by the software. Figure 1 illustrates this file structure. defines ATR SC traffic processed ATR SC TRADAS-ATR TRADAS-SC Figure 1: The traffic directory structure 2

4 After creating the directory tree as shown in Figure 1, the next step is to copy station definition files to the corresponding directories. To run the ATR-TDA2 program, the user must first place at least one ATR station definition file in the directory, traffic\defines\atr\, which defines the detector IDs for each station. For example, the ATR definition files in the directory may look like: traffic\defines\atr\ ATRDets txt ATRDef txt Test_ATRDef txt Each ATR definition files should include a string ATRDets or ATRDef in the name and inclusion of the created date is recommended as shown in the above examples. The user is required to select one of the available definition files using Settings. Similarly, the SC-TDA2 program also requires a short-duration count (SC) station definition file in the directory, traffic\defines\sc\, for example, the definition files in the directory may look like: traffic\defines\sc\ SCDets txt SCDef txt Test_SCDef txt The SC definition files should include a string SCDets or SCDef in its name. User is required to select one of the definition files before running SC-TDA2. Important Note: As of Dec 20, 2017, ATR-TDA and SC-TDA programs were upgraded to ATR-TDA2 and SC-TDA2 programs. These new versions no longer require the archives generated by ArcCon or ArcCon_Rochester programs and directly obtain traffic data from the RTMC IRIS server through on-line access. 3

5 2. ATR-TDA2 2.1 Brief Implementation Description Automatic Traffic Recorder (ATR) data is a continuous-count traffic data recorded every hour, 24/7, all year, non-stop. The ATR-TDA2 application computes ATR data from the Regional Traffic Management Center (RTMC) traffic data (mostly from loop detectors but there are some radar detectors). For generating the continuouscount data from RTMC data, the basic algorithm imputes missing data based on spatial and temporal inferences of traffic data. The equivalent spatial relation is created by defining primary, secondary, and tertiary detector sets, based on an equivalent traffic flow. Among the three sets of detector data, the program chooses the data set with least missing for computing the ATR data. Temporal relation is created based on the trend that the same day-of-week volume patterns in neighboring weeks are similar. The mathematical description and the detailed algorithm of the imputation process is available in the final report: TMC Traffic Data Automation for Mn/DOT s Traffic Monitoring Program, Minnesota Department of Transportation, Report No. MN-RC , July, The implementation of the software is described for non-programmers, and thus it is focused on description of procedural steps. The following list shows the steps of the ATR computation from 30 second detector data available from RTMC. 1. Load the station definition file specified in Settings. 2. Read in the Last Run Ending date from the Settings. 3. Generate an array of consecutive dates consisting of starting from Monday and ending on Sunday since the Last Run Ending date. 4. Check if the traffic data for the computing dates in the array are available. If data for the entire week is not available, give an error message and abort the computation. 5. Read in the 30 second data for primary, secondary, and tertiary detector set. 6. Compute the missing percentage of each detector set and choose the set with least missing. Spatial imputation. 7. Impute the 30 second detector data for only randomly missing data patterns using the Non-normal Bayesian Linear Regression (NBLR) algorithm, i.e., temporal imputation. 8. Compute the 30 second station data using the imputed detector data. 9. Pack the 30 second station data into 5 minute station data. 10. Impute the 5 minute station data using NBLR algorithm only for the randomly missing patterns. 4

6 11. Pack the imputed 5 minute data into hourly data. 12. For hourly missing data, impute them using the historic imputation algorithm if the option for historic imputation was selected in the Settings. 13. Output the data in an ASCII file format defined by the Office of Mn/DOT TDA. The historic imputation algorithm applied is briefly described. When a large amount of data is missing such as many hours or the whole day, the hourly data is copied from the historic database which is regularly updated whenever a date with no missing data is available. The historic database does not use holidays nor near holidays for the update of the database. Imputation based on the historic database of the detector data is referred to as the historic imputation. Historic data of each hour per day of the week is created using the following relation. Let d(n) be a good data of date n, then the historic data is recursively updated using the exponential moving average: d(n+1) = d(n)*0.5 + d(n-1)*0.5 and the data d(n+1) is kept in the database. This approach assigns a higher weight on the most recent data since d(n-1) was recursively produced using past data. The weight for the averaging is exponentially decreased as the dates get old. 2.2 How to Use the ATR-TDA2 Software In order to run ATR-TDA2, an ATR station definition file must be present and selected. If an ATR definition file was not selected or not present, all processing buttons are automatically disabled so that the user cannot run the ATR-TDA2 program. To select an ATR definition file, press the Settings\Parameters menu which should bring up the Parameter Settings window as shown in Figure 3. The Traffic Data Root Directory is the root node, traffic directory in Figure 1 and should be entered using the Browse Folder button in order avoid spelling mistakes. The textbox below the root directory shows the directory that is accessed by ATR-TDA2, derived from the traffic root directory. The check mark Perform historic imputation based past good data should be checked if historic imputation is desired to be used. The textbox under ATR Station Definition File should have the user selected ATR Definition file. It should be set using the button provided. When the ATR-TDA2 produces ATR data from RTMC traffic data, it can simultaneously produce two formats: the legacy input format for a SAS program and the FHWA volume format for TRADAS. The user is responsible to set the FHWA Export Folder using the Browse button. The ATR-TDA2 will then send all FHWA volume files to this directory. The setting parameters must then be saved using the Save button, otherwise they will be lost. 5

7 Figure 2: ATR-TDA Parameter Settings window Figure 3 shows the main window of ATR-TDA2. There are two options of computing the ATR files, Manual Run or Auto Run. The Manual Run is used to run one week at a time. To use this function, click any day of the week from the calendar. The software will then automatically determine one week (starting from Monday and ending on Sunday) that includes the date the user selected and produce the ATR data for the week. The second option is the Auto Run. This option computes the ATR files from the week after the Last Ending Date to the most recent week. This Auto Run function can be executed from the Windows Scheduler. Configuration of this option is described later. When the check mark Generate FHWA VOL file when ATR files are created is checked, the ATR-TDA2 produces both the regular ATR formatted data as well as the FHWA volume formatted data, which are then saved to the respective directory. The Generate FHWA Vol file from Existing ATR Files button is used when an FHWA formatted volume files need to be produced from ATR files. Clicking this button opens a folder dialog in the ATR data directory from which the user can select multiple ATR files for conversion. 6

8 Figure 3: ATR-TDA2 computation tab In order to automatically run the ATR-TDA program at the scheduled time every day, use the following setup for Windows Task Scheduler. 1. Open Windows Task Scheduler by Control Panel/Scheduled Tasks. 2. Double click on Add Scheduled Task. 3. Click the Browse button and browse to select C:\Program Files\Bulldog\ATR- TDA\ATR-TDA2.exe. 4. Select the Every Day option and set the Start time as desired. (you can also set Once a Week option) 5. Enter User Name and Password 6. Click Finish. 7. In the Action, select Start a program and select C:\Program Files\Bulldog\ATR- TDA\ATR-TDA2.exe for the Program/Script. In the Add arguments (optional) textbox, enter -auto. 7

9 2.3 Plot Utility of ATR-TDA2 The second tab named Plot Utility (shown in Figure 4) provides visualization of the ATR data computed. To use this utility, browse and load any ATR data files produced by the ATR-TDA2 program. Next, select ATR stations and dates, and then press one of the three buttons available. An example of Hourly Plot is shown in Figure 5. Figure 4: ATR-TDA2 Plot Utility tab Figure 5: An example of hourly plot 8

10 The daily plot shows daily volume of stations using a bar graph but it can be changed into a line graph using the Edit button. The data plotted can be retrieved using Excel by pressing the Export to a CSV file button which should open up data in Excel, from which data may be saved. 2.4 ATR Data and Log File Format ATR data outputs produced by the ATR-TDA2 program are two text files: ATR data file and log file. The filenames conform to the format where the date is the ending date of the week, i.e., ATRyyyymmddw1.dat ATRyyyymmddw1.log In the filename, w1 indicates that the file contains only one week of ATR data and the format is described next. Format of ATRyyyymmddw1.dat: All characters in the file are ASCII characters. One day data of a station in one direction occupies two rows: 12 hours per each row corresponding to AM in the first row and PM in the second row of the day. The format of a single line is summarized in Table 1. Table 1: Field Description of ATR Data File Digit Number Description Position of Digits 1 1 Always AM=1, PM= Month, Day of the month, Last two digit of the year 9 1 Day of the Week: Sun=1, Mon=2, Tue=3, Wed=4, Thu=5, Fri=6, Sat= Station ID 13 1 Lane direction of the station, E,W,S,N,R A Set of five digits represents the hourly volume. Sixty digits (12*5=60) are consecutively concatenated in the order representing hours 1 st to 12 th depending on AM or PM. 9

11 Interpretation of data by an example Below was taken from top four rows of an actual ATR file, ATR W1.dat E E W W Interpretation of the first line: E Digit Value Meaning Position 1 2 ATR data, always AM January st day Year Day of the Week: Monday Station ID = E Lane Direction, East A set of five digits representing the hourly volume. The data consists twelve five digits. ATR Log-File Data Format: The log file consists of two sections. The first section shows the statistics of missing detectors in the original traffic data and the missing percent of the station. Each set of these statistics is specified for Primary (P), Secondary (S), and Tertiary (T). Let s take an example. > Inspecting missing det files and missing-data (MD) on Tuesday, August 04, :: P: None, MD=.0%: S: None, MD=.0%: T: None, MD=.0% 301-7:: P: None, MD=.0%: S: None, MD=.0%: T: None, MD=.0% 303-1:: P: None, MD=.0%: S: None, MD=.0%: T: None, MD=.0% 303-5:: P: None, MD=.0%: S: None, MD=.0%: T: None, MD=.0% 305-1:: P: None, MD=.1%: S: None, MD=.0%: T: 0,MD=100.0% 305-5:: P: None, MD=.0%: S: None, MD=.0%: T: 0,MD=100.0% 309-1:: P: 341,342,343,MD=100.0%: S: None, MD=.0%: T: 335,336,337,338,-339,-340,MD=100.0% 309-5:: P: 178,179,189,MD=100.0%: S: 95,MD=25.0%: T: -181,-182,183,184,185,186,MD=100.0% 315-1:: P: 494,495,496,MD=100.0%: S: 266,267,525,1038,MD=100.0%: T: 156,268,269,270,533,MD=100.0% 315-5:: P: 256,257,1003,MD=100.0%: S: 1006,1007,1008,MD=100.0%: T: 110,254,255,1002,MD=100.0% 10

12 Line-1: 301-3:: P: None, MD=.0%: S: None, MD=.0%: T: None, MD=.0% Interpretation: The station ID is 301 and the direction is 3. Primary (P), secondary (S), tertiary (T) had no missing detectors. MD=.0% means that the missing data percentage is 0%. This also means no imputation is needed. Line-7: 309-1:: P: 341,342,343,MD=100.0%: S: None, MD=.0%: T: 335,336,337,338,-339,- 340,MD=100.0% Interpretation: The station ID is 309 and the direction is 1. The missing Primary detector IDs from the traffic file are 341, 342,343 and the missing data percentage is 100%. The secondary detectors are 100% available from the traffic file. The tertiary detectors are missing 100% from the traffic file, and the detector IDs are 335,336,337,338,-339, and The detector IDs with a negative sign indicate subtraction of the detector volume from the total station volume. In this case, the ATR- TDA used the secondary detector set. In the second section, the log data shows imputation adjustment details. The first line shows the station ID and direction, total volume of the day, and the percentage of imputation adjusted. The remaining part shows the hourly volume of 24 hours. For each hourly volume, it shows which detector set was used (P, S, or T), what was the volume before imputation (raw data), missing percent of raw data, and the amount of volume adjusted by imputation. The hourly volume follows the format described below: [P, S, T, or B and volume before imputation]:[missing percent]:[the amount adjusted] The letter B indicates that temporal and/or historic imputation occurred. Let s take an example dailyvol=11427 ImpAdj=33.85% P3:.0:0 P1:.0:0 P0:.0:0 P1:.0:0 P7:.0:0 P70:.0:0 P645:.0:0 P2057:.0:0 P1838:.0:0 P862:.0:0 P345:.0:0 P224:.0:0 B243:2.2:-57 B82:66.7:-4 B56:66.7:320 B228:66.7:413 B254:66.7:830 B263:66.7:987 B190:66.7:541 B117:66.7:147 B27:66.7:171 B31:66.7:269 B8:66.7:169 B7:66.7:82 Line 1: dailyvol=11427 ImpAdj=33.85% Station ID is 327-3; the daily total volume was 11,427; and imputation volume adjustment was 33.85% Line 2: P3:.0:0 P1:.0:0 P0:.0:0 P1:.0:0 P7:.0:0 P70:.0:0 Hour 0 Hour 1 Hour 2 Hour 3 Hour 4 Hour 5 Selected Set P P P P P P Raw Volume Missing % Adjusted

13 Line 3: P645:.0:0 P2057:.0:0 P1838:.0:0 P862:.0:0 P345:.0:0 P224:.0:0 Hour 6 Hour 7 Hour 8 Hour 9 Hour 10 Hour 11 Selected Set P P P P P P Raw Volume Missing % Adjusted Line 4: B243:2.2:-57 B82:66.7:-4 B56:66.7:320 B228:66.7:413 B254:66.7:830 B263:66.7:987 Hour 12 Hour 13 Hour 14 Hour 15 Hour 16 Hour 17 Selected Set B B B B B B Raw Volume Missing % Adjusted Line 5: B190:66.7:541 B117:66.7:147 B27:66.7:171 B31:66.7:269 B8:66.7:169 B7:66.7:82 Hour 18 Hour 19 Hour 20 Hour 21 Hour 22 Hour 23 Selected Set B B B B B B Raw Volume Missing % Adjusted In summary, the hourly volumes for the above 5 lines of log data indicate: Hour Before Imp Missing % After Imp Hour Before Imp Missing % After Imp From this data, we can learn that two thirds of detectors did not work during the 12nd 23 rd hours. After imputation, the afternoon peak hour volumes were increased to a normal range. 12

14 3. SC-TDA2 3.1 Brief Implementation Description SC-TDA2 computes Short-Duration Count (SC) data for the defined stations using the RTMC traffic data. For missing data, it applies similar imputation procedures as described in the ATR-TDA2 program. The main difference is that the SC-TDA2 program does not apply corrections based on historic imputation. The implementation of software is described for non-programmers, and thus it is only focused on description of procedures. The following list shows the overall steps of the computation. 1. Load the most recent station definition file. 2. Compute missing statistics of primary, secondary, and tertiary detector set and choose the set with least missing data for each station 3. Compute hourly data for each day using the following steps for each station: a. Impute the 30 second detector data for only randomly missing data patterns using the Non-normal Bayesian Linear Regression (NBLR) algorithm. b. Pack the 30 second detector data into station data. c. Pack the 30 second station data into 5 minute station data. d. Impute the 5 minute station data using NBLR algorithm only for the randomly missing patterns. e. Pack the imputed 5 minute station data into hourly data. f. If a single hour missing data exists, impute it by averaging the neighboring hours 4. Pack hourly data of all stations into a single binary file for each day. This produces 365 binary files since the period is normally selected for one year. 5. Perform week-to-week imputation if the option is chosen. This imputation uses the volume trend of neighboring weeks of the same day. Holidays and near holidays are excluded from the inference of data. 6. Pack the hourly data for all stations into a single binary file and save it as Hryyyymm.yr or Hryyyy-mm.yrImp, depending on whether it processed week-to-week imputation or not. 7. Pack the hourly volume data into daily volume data and save the whole data to a single binary file with the filename formatted with stayyyy-mm.days where yyyymm is the ending month of the period. 8. Search the consecutive zeros lasting longer than 7 days (one week) and mask out as missing data. 9. Search hourly volume for the highest hour, 30 th high hour, and 500 th high hour, and then output the data into the filename formatted as HighHouryyyy.txt. 13

15 10. Filter partial day data. Compute daily average of a station by averaging, excluding missing data. If a single day volume is less than 20% of the average, consider it as a partial day data and remove from the AADT computation. 11. Compute AADT using either by the AASHTO method or by simple averaging, depending on the option chosen by the user and save the result in a file with its name formatted with ADTSampleyyyyAASHTO.txt or ADTSampleyyyy.txt, depending on the AADT computation option chosen. The AASHTO AADT is computed using the following steps: 1. Compute day-of-week (DOW) ADT for each month, i.e., average volume of Sunday, Monday, Tuesday, Wednesday, Thursday, Friday, and Saturday of the month. This produces seven values for each month. Call these values ADT_DOW (i,j) where i=1,2,,12 and j=1,2,, Compute Avg_ADT_DOW(j) = [ i=1 ADT_DOW (i, j) ]/12 for j=1,2,, AASHTO_AADT = [ j=1 Avg_ADT_DOW(j) ]/ How To Use the SC-TDA2 Program The software consists of three tabs, and first two tabs are used for SC data generation (Figure 6). The third tab is only used to retrieve hourly or daily volume data after the computation is completed. As the prerequisite, a SC station definition file must be set from Settings. For generating the SC data, please follow the steps provided below. 1. If the Traffic Data Root Directory is not set, set it using the Browse Folder button of the Settings/Parameters menu. 2. Set the Beginning Date and Ending Date of the computing year. 3. Compute the hourly data of each station by clicking the Generate Daily Vol Files button. This step takes a long time (4 to 5 hours for the entire TC network) since it is processing the data for the entire year for all stations, so the user should let it run overnight until the processing is completed and should not close the program in the middle or processing. 14

16 Figure 6: SC-TDA2 screen Below describes the folder and files produced by selecting the period 1/1/ /31/2010. File or folder HrSta\ Description This folder contains hourly volume of all stations per each day in a binary format, i.e., each file contains hourly data (24 values) for all stations for the date specified in the filename. ADTSample2010AASHTO.txt The final AADT for each station computed from the selected year. The data is stored in a text format. The word AASHTO in the filename does not present if it is a simple averaging AADT. HighHour2010.txt Hr yr Hr yrImp Sta days Hourly volume for the highest hour (1 st ), 30 th high hour, and 500 th high hour. The data is stored in a text format. This contains a table of the individual files in the folder HrSta. The data is stored in a binary format. This file is same as the Hr yr file except that the week-to-week imputation method was applied. This file contains daily volume of each station for the entire year. The data is stored in a binary format. 15

17 Change the tab to the AADT Computation tab (shown in Figure 6). 1. Load the directory traffic\processed\sc\yyyy-mm using the Browse Folder button where yyyy denotes the ending year and mm denotes the ending month of the selected period. 2. Select or deselect the option for week-to-week imputation 3. Select or deselect the option for AADT computation using the AASHTO AADT method or simple averaging. 4. Choose an option for producing TRADAS export files. If this option is selected, the TRADAS Export Directory must be selected using the Browse Folder button. 5. Press the Compute All button. It will then compute and produce the high-hour rank, AADT, and TRADAS export data files along with hourly and daily traffic volume files in the yyyy-mm directory. ***Note: The directories are automatically set to the most recent computation by the software. Consequently, the user just needs to accept the default directory set by the program and press the Compute All button. It may not have the proper directories set if the program was closed in between and reran. Figure 7: SC-TDA AADT Computation Tab 16

18 3.3 Output File Data Formats Although multiple intermediate data files are produced, the final SC outputs of the SC-TDA2 program are two text files that follow the format specified by the Mn/DOT TDA office, i.e., AADT data and hourly order-statistic data. AADT Data File The AADT data is contained in the filename with ADTSampleyyyyAASHTO.txt where yyyy indicates the year computed. It is a text file and consists of five columns separated by comma. The columns are: Column Description number 1 Station ID (Sequence number) 2 Direction code: 1=N, 2=NE, 3=E, 4=SE, 5=S, 6=SW, 7=W, 8=NW, 9=N- S or NE-SW combined, 0=E-W or SE-NW combined. 3 Ending date, MM/dd/yyyy where MM=month, dd=day, yyyy=year 4 AADT of the station (directional) 5 Comment enclosed by double quotations ***. It contains the number of valid days available for computation, data source (TMC), a list of months with no complete week data. Example) The following line show an example of a single row in the AADT data file , 1, 12/31/2011, 31395, "144 TMC " This line is interpreted as: station_id=10974, Lane_direction=North, Ending_Date=12/31/2011, AADT=31395, and the comment which says 144 days were available for AADT computation and the months that did not have at least one complete week are 7, 8, 9, 10, 11, and

19 Hourly Order-Statistic File The HighHouryyyy.txt file contains ordered hourly volume data of 1 st, 30 th, and 500 th highest rank volume. It consists of four columns by following the format described below. Column Description Number 1 Station ID (Sequence number) 2 1 st highest hour: date and hour, the hour volume, directional distribution 3 30 th highest hour: date and hour, the hour volume, directional distribution th highest hour: date and hour, the hour volume, directional distribution Example) The following line show an example of a single row of the hourly high volume ranked data. 9835, 08/29/2011: /(3+7)=0.52, 08/19/2011: /(3+7)=0.54, 01/21/2011: /(3+7)=0.52 It is interpreted as: Station ID = 9835; The highest hourly volume = 11,866 which occurred on 8/29/2011 at 4:00pm, East directional volume was 52%; The 30 th highest hourly volume = 11,316 which occurred on 8/19/2011 at 4:00pm, East directional volume was 54%; The 500 th highest hourly volume = 10,056 which occurred on 1/21/2011 at 4:00pm, East directional volume was 52%; 18

20 TRADAS AADT file format The AADT data is produced to export to TRADAS. The filename follows the following format. AADT-TRADAS-yyyymmdd(begin date)-yyyymmdd(end date).txt. An example filename is AADT-TRADAS txt Inside the file is a text data, consisting of eight columns separated by comma. The columns are: Column Description number 1 V24 2 Station ID (Sequence number) 3 Direction code: 1=N, 2=NE, 3=E, 4=SE, 5=S, 6=SW, 7=W, 8=NW, 9=N- S or NE-SW combined, 0=E-W or SE-NW combined. 4 Beginning date, MM/dd/yyyy where MM=month, dd=day, yyyy=year 5 Ending date, MM/dd/yyyy where MM=month, dd=day, yyyy=year 6 AADT of the station (directional) 7 It contains the number of valid days available for computation, a list of months with incomplete week data separated by space. 8 Data Source, i.e., RTMC Example) The following line show an example of a single row in the AADT data file. V24,98357,01/01/2012,12/31/2012,53914,255 days ,RTMC FHWA Volume Format It contains hourly data for each station for the period defined, typically one year. It follows the standard format defined by Traffic Monitoring Guide, May 1, There are some default values set, which are described including variables in the following table. Column(s) Description 1 Record Type= 3, Traffic volume record 2-3 FIPS State Code= 27, Minnesota 4-5 Functional Classification Code= Station Identification=SC Sequence number in 6 digites 12 Direction of Travel Code=SC Station Direction 13 Lane of Travel= 0, combined lanes Year of Data=yy 19

21 16-17 Month of Data=MM Day of Data=dd 20 Day of Week (1=Sun, 2=Mon, 3=Tue, 4=Wed, 5=Thr, 6=Fri, 7=Sat) 21-25,, Traffic Volume Counted Field=(5digit hourly vol) * Restrictions= 0, no restrictions 3.4 Data Plot Tab When SC-TDA2 runs, it not only produces the final AADT data but also produces hourly (just like ATR) and daily traffic data for the entire year. You can use the Data Plot tab to retrieve hourly and daily traffic volumes in three forms: text, Excel, or graph. This tool is useful when you want to see hourly volumes of a day or daily volumes of entire year. A screen shot of this Tab is shown in Figure 8. Left side of the window is used for hourly volume retrieval, and the right side is used for retrieval of daily volume. For retrieving hourly volume of a day, an hourly volume file with.yr or.yrimp extension should be loaded using the Browse/Load button, which would populate the Stations and Dates ListBox. The user may then select one or more stations and a date. The three buttons on the right side of the dates ListBox provide retrieval options of text, Excel, or graph. Figure 9 shows an example of the hourly plot of four stations selected as shown in Figure 8. Daily volume is retrieved using the right side plane. Loading a daily traffic volume file ( *.days ) populates the station list. The text, Excel, and graph outputs, all of them, provide daily volume for the entire year. An example plot is shown in Figure 9. 20

22 Figure 8: Data Plot tab Figure 9: Hourly volume plot of four stations selected 21

23 Figure 10: Monthly volume plot of two stations 22

24 4. Concluding Remark Although the outside interfaces of the programs described in this manual show simplicity, the internal programs are actually very complex, developed by multiple years of research and development, writing thousands of lines of codes. The mathematical procedures applied for imputation may be understood at the level of a Maters of Science (M.S.) in Engineering or Mathematics. If anyone wishes to explore the mathematical procedures used in the software for multiple imputation, reading of the following two books are highly recommended. [1] Rubin, Donald B., Multiple Imputation For Non-Response in Surveys, Wiley Series in Probability and Mathematical Statistics, [2] Little, R. J. A. and D. B. Rubin, Statistical Analysis with Missing Data, Wiley Series in Probability and Mathematical Statistics, The software packages will be updated regularly, typically triggered by the following events (1) error was found, (2) operating system (OS) versions have been changed, (3) RTMC changed the data format, or (4) after a major upgrade. Presently, the OS supported by the software are Windows 7 and 10 and it requires.net Framework 4.5. The latest versions of the software are distributed through the web site: Lastly, I would like to thank three people contributed towards this work: Mark Flinner, Gene Hicks, and Megan Forbes. Their inputs were invaluable for research, development, and implementation of the final software packages. 23