THIS PRESENTATION IS A MODIFIED “REBROADCAST” Originally a preliminary presentation prior to a work session to review RoadMAP performance guidelines It has been reformatted to fit your screen! The books handed out are a companion piece to this presentation (I HAVE NO BOOKS)
SO WHAT IS ROADMAP??? Road Maintenance Accountability Program ALDOT’s new MMS implemented July/August CitiTech Systems V Approximately 300 licensed users generate 100,000 work reports annually for the 70 routine activities that are performed. Has accumulated some interesting data….BUT, generating reports and analyzing data within RoadMAP has proven to be…..challenging (to say the least). Performance speed and credibility issues with end users affecting acceptance of program. Hence, the need for additional tools.
METHOD OF ANALYSIS - SPSS Statistical Package for the Social Sciences Now owned by IBM ALDOT licenses SPSS Modeler Purchased fall 2012 Training provided by IBM in December 2012 Data mining & analysis began Spring 2013
SPSS STREAM TO READ ROADMAP DATABASE AND PRODUCE ANALYSIS OUTPUT FILE
Data Prep Data QA Data Selection Output
HOW WE USED WHAT WE FOUND IN IBM SPSS Annual Maintenance Management Meeting, August 8-9, 2013 Breakout sessions (1 ½ hours in length) Attendance required for Division Engineers, Division Maintenance Engineers and their staff, District Managers and anyone else we could pull off the street. Each attendee was provided book of data mining results for a group of activities.
MATERIALS FOR TAC BREAKOUT SESSIONS Technical Advisory Committee (TAC) procedures Actual Average Daily Production (ADP) and Average Reported Crew Sizes 12 months of RoadMAP work reports: July 1, 2012 to June 30, 2013 Prior to conversion of District 81 to District 56. Prior to establishment of Southwest Region Three Groups of Activities Same as last year (some additions, no deletions) One handout per Group of Activities Assignments Group 1:Divisions 3, 4 and 9 Group 2:Divisions 1, 5 and 8 Group 3:Divisions 2, 6 and 7
“RULES OF ENGAGEMENT” Attend breakout session as noted in agenda Session moderator will identify activities to be reviewed Activities will be reviewed as time allows Focus will be on activities with greatest use and/or greatest expenditures Prior year modifications (if any) will be noted by the moderator before beginning discussion on an activity Voting Process One vote per Division/Area. Division/Region Engineer designates voting member. Proposed changes require a majority vote to be moved forward. Summary of each proposed change, the breakout discussion and the vote will be provided during the general session.
Activity Number and Description Crew Size Average Daily Production Group Number
2 1 Three Items Analyzed Actual ADP Crew Size Crew Hours Six Pieces of Information for Each Item A.Count: Number of Work Reports for this Activity B.Mean: Average of All Values C.Min: Lowest Value D.Max: Largest Value E.Median: “Middle” Value F.Mode: Most Common Value Are adjustments needed? Crew Size Actual ADP A. B. C. D. E. F. A. B. C. D. E. F. A. B. C. D. E. F.
APPLYING THE MEAN, MEDIAN AND MODE Median – arrange the values in order and find the one that is right in the middle. Half of the values are larger than the median and half are smaller. Mode – the value that shows up the most often. The mean is more “trustworthy” when both the median and mode are “close” to it. Why? When large “outlier” values are present (entry errors, for example), the mode is generally unchanged and the median changes slightly but the mean changes a great deal. When these three numbers are close to each other, it is an indication that large outliers are not present (or were removed). When the mode is exactly equal to the planned ADP shown on the Performance Guideline, it may indicate that actual accomplishments are often not being entered (i.e., default value is being accepted).
DISTRIBUTION OF ACTUAL ADP – ACTIVITY 6020 (MAJOR PATCHING) ALL DIVISIONS – ALL VALUES Mean = 375 tons Median = 80 tons Mode = 0 tons Max > 125,000
DISTRIBUTION OF ACTUAL ADP – ACTIVITY 6020 ALL DIVISIONS – EXTREME OUTLIER REMOVED Mean = 100 tons Median = 78 tons Mode = 0 tons Max < 800
DISTRIBUTION OF ACTUAL ADP – ACTIVITY 6020 ALL DIVISIONS – EXTREME OUTLIER REMOVED (DIFFERENT THRESHOLD) Max < 300
DISTRIBUTION OF MAINTENANCE UNIT COST – ACTIVITY 6382 CABLE RAIL MAINTENANCE
DISTRIBUTION OF MAINTENANCE UNIT COST – ACTIVITY 6381 GUARD RAIL MAINTENANCE
ACTUAL ADP – 6252 MOWING (NON-INTERSTATE) (BY DIVISION AND ROADCLASS) Road ClassDivision Error in Road Class?
AVERAGE CREW SIZE MOWING (NON-INTERSTATE) (BY DIVISION AND ROADCLASS) Road ClassDivision Error in Road Class?
Dots represent Divisions AVERAGE ACTUAL ADP VS. AVERAGE ACTUAL CREW SIZE BY DIVISION – 6252 MOWING (NON-INTERSTATE) Different daily production for same crew sizes Daily production increased with larger crew sizes
AVERAGE ACTUAL ADP VS. AVERAGE ACTUAL CREW SIZE BY DIVISION, DISTRICT AND ROADCLASS – 6252 MOWING (NON-INTERSTATE) Dots represent Districts Daily production and crew sizes consistent Different daily production for same crew sizes Daily production increased with larger crew sizes
SO HOW ARE END USERS BENEFITTING??? TAC approved changing ADP on 4 activities & crew sizes on 4 activities Compare productivity of districts for specific activities. Compare number of work reports for a specific activity for each district.
SO HOW ARE END USERS BENEFITTING??? TAC approved changing ADP on 4 activities & crew sizes on 4 activities Compare number of work reports for a specific activity for each district. Compare productivity of districts for specific activities. One DME discovered a district with heavy emphasis on litter pickup and major patching activities at expense of herbicide spraying (weedy roadsides) and sign installation & maintenance (pallets of new signs at district yard not installed). RESULT: Requested additional resurfacing projects (to reduce patching efforts), will begin contracting litter pickup along with ALL mowing, ordered more emphasis on herbicide application and REASSIGNED former sign inspector to Construction section. Hunger for deeper understanding of the data as they look to optimize their resources. ($$= okay, staffing = not okay)