Statistical Analysis of Road Traffic Data

Y

yasharif

I have a database:
Interstate ADT(avg daily traffic) IRL 1yrRL 2ndyrRL...5thyrRL
I-95 15000-25000 800 750 650 340

I-95 is a highway where average daily traffic is from 15000-25000. The roadmarking or roadsign that were put initially to guide the car/driver at nighttime had IRL(initial retro reflective number) of 800 and the material was polyurea. It can be thermo or paint also. Then an instrument is used and the reading of the roadmarking taken a year later and found 750 and then the later years 650,340...and son on may be upto 7th year.

I developed an exponential degradation model from the database but does not look satisfactory to my BOSS.

Now my goal is to find the cost effectiveness of these materials.

Is there any otherway you can suggest or advice me..

Thank you once again.

Regards
Sharif
 

Steve Prevette

Deming Disciple
Leader
Super Moderator
Seems like a fairly straight forward analysis. Is the IRL a linear, measured quantity, or a subjective scoring?
 
Y

yasharif

No it does not seem to me linear.

There are 3 materials for roadway marking : polyurea, thermo and paint.

The first marking that is put is measured and that is called initial retroreflective number. And then the other readings are taken at 6 months, 1 yr, 2 yr, 3 yr, 4yr and 5 yr.

The minimum retroreflective number to be valid for the materials are with a standard number.

Polyurea (white): 800
Polyurea (yellow): 500

Thermo (white): 375
Thermo (yellow): 250

Paint (white): 225
Paint (yellow): 200



Now a sample :

US 17 route: Polyurea

(initial) (6month) (1yr) (2yr) (3yr) (4yr)
White: (629) (490) (478) (368) (324) (No data available)
Yellow: (517) (450) (390) (326) (272) (No data available)


there are several routes like this for intertate, primary and secondary.

This is just a sample.

I am trying to find an equation (linear/exponential) that will predict the retroreflective number after 10 years.

Any suggestions will be appreciated.

Thank you very much.

Regards
Sharif
 
P

PaulJSmith

I suspect there is more to consider in your equation; such as how each of those materials are affected by environment (heat, cold, moisture, sunlight, etc.). You may need data from many different areas, which may give you results that vary by region/climate. I doubt it will be as simple as using just one generic formula.
 
Y

yasharif

Thanks for your reply.

I know that it might have the temperature and environmental effect but still trying to avoid those parameters to make the equation simplified.

What would be better approach?

Thanks.
 

Steve Prevette

Deming Disciple
Leader
Super Moderator
How many different combinations of location, material, and other factors are there? Are there enough replications in a given data set to do a non-parametric analysis (the score changes by this much first year, second year, etc).

If you are talking US 17 and I-95, you must be in the Southeast . . . I live in South Carolina. Are the data in some sort of database or spreadsheet?
 

Steve Prevette

Deming Disciple
Leader
Super Moderator
Also, how is the IRL measured? What does the number mean? Are we talking road signs or highway markings or both? Is the road sign have surface grime removed from it prior to taking the reading?

I assume that IRL once lost cannot be recovered (a later year's measurement can never be higher than the previous year's), correct?

Before these data were collected, was there any plan as to how to analyze the resulting data?
 
Y

yasharif

RL-retroreflective numbers are measured through an instrument.

The road surface asphalt/concrete are made clear and then the material(Polyurea/ Thermo/ Paint) is put at 40F temperature on the Road Surface to guide the driver.

The number basically means for instance HOW MUCH REFLECTION THE MATERIAL CAN create to the drivers eye for the visibility in day and specially at night time.(Like as right arrows, left arrows, straight arrows, edge line to prevent the driver to go off the road , yellow centre line to keep the 2 way system separate and consistent..etc)

Yes the number starts getting down after the road surface is being painted with polyurea/thermo or paint due to specially cars tire abrasion and many more things. They cant be recovered unless the road surface is painted with those material again.

No, there was no plan to analyze the data before collecting these data. I wish a DOE was done before measuring these RL. If there was a design matrix it would have been easier.

Do you have any suggestion?

Any mathematical/statistical model might give me a solution.

I am basically trying to figure out the cost/benefit ratio of those materials.

But before that I need a model to introduce which can predict the final RL after a certain year which will give me the result for cost analysis of those 3 materials..

Thank you very much for your quick reply.

Regards
Sharif
 

Steve Prevette

Deming Disciple
Leader
Super Moderator
Without seeing the entirety of the data it is hard to see what to do, but here (in the blind) is what I would do:

For each family of data, calculate the percent reduction in the score from time 0 to year 1; year 1 to year 2; year 2 to year 3; etc. Then I would start examining if the error within a group fits the binomial distribution (p-chart control limits). Then look across the years to determine if the percentage reduction changes year to year. If all of these subgroups of percentage reductions are within binomial limits, then that would say there is a predictable reduction each year.

If the percent reduction is NOT stable year to year, then start looking across the different families of data to see if there is a consistent pattern to the percent reductions.

Depending on how all this goes, one should be able to come up with a IRL reduction curve versus time, with confidence limits and prediction limits.
 
Y

yasharif

Thanks for your quick reply.

I used to live in columbia,sc. I completed my graduation and moved to NC.

Yes, there are lots of parameters.

I am just giving one snapshot of the data I am considering for analysis. Its a huge large database.

I am trying to simplify the database with lowest parameters like ADT, Material color and Initial RL.

Problem is , the party who was engaged in collecting or measuring the data dint do their job properly. They did not collect the data properly. Sometimes they did not collect the data at the beginning and all on a sudden they are showing the data of Rl for a year later and onwards.

Although the chart has shows the column upto 9yrs but they only have data upto 4 or 3 yrs...!!

I developed an exponential equation but still was looking for something better or may be a new approach to model or develop an equation.

Thanks.
 

Attachments

  • snapshot.png
    snapshot.png
    4.7 KB · Views: 182
Top Bottom