Mobile QR Code QR CODE : Journal of the Korean Society of Civil Engineers

  1. *ํŽœ์‹ค๋ฒ ๋‹ˆ์•„ ์ฃผ๋ฆฝ๋Œ€ํ•™๊ต (*The Pennsylvania State University)


์‚ฌ๊ณ ๋ชจํ˜•, ํ™•๋ฅ ์  ๋ชจ์ˆ˜๋ฅผ ๊ณ ๋ คํ•œ ์Œ์ดํ•ญํšŒ๊ท€๋ถ„์„, ๊ณ ์†๋„๋กœ ๊ธฐํ•˜๊ตฌ์กฐ, ์ด์งˆ์„ฑ
Accident model, Random parameter negative binomial regression analysis, Interstate highway geometrics, Heterogeneity

  • 1. ์„œ ๋ก 

  •   1.1 ์—ฐ๊ตฌ์˜ ๋ฐฐ๊ฒฝ ๋ฐ ๋ชฉ์ 

  •   1.2 ์—ฐ๊ตฌ์˜ ๋‚ด์šฉ

  •   1.3 ์—ฐ๊ตฌ์˜ ๋ฒ”์œ„

  • 2. ๊ธฐ์กด ๋ฌธํ—Œ ๊ณ ์ฐฐ

  •   2.1 ๊ตํ†ต์‚ฌ๊ณ  ๋ชจํ˜•

  •   2.2 ๊ตํ†ต์‚ฌ๊ณ ์™€ ๊ธฐํ•˜๊ตฌ์กฐ์™€์˜ ๊ด€๊ณ„

  • 3. ์ž๋ฃŒ๊ตฌ์ถ•

  •   3.1 ๋ถ„์„๋ฒ”์œ„

  •   3.2 ๊ฐœ๋ณ„์ž๋ฃŒ๋ถ„์„

  • 4. ๋ชจํ˜•์˜ ๊ตฌ์ถ• ๋ฐ ๊ฒฐ๊ณผ๋ถ„์„

  •   4.1 ํ™•๋ฅ ์  ๋ชจ์ˆ˜๋ฅผ ๊ณ ๋ คํ•œ ์Œ์ดํ•ญ๋ชจํ˜• (Random Parameter Negative Binomial)

  •   4.2 ๊ฒฐ๊ณผ๋ถ„์„

  •   4.3 ๋ชจํ˜•๋น„๊ต

  • 5. ๊ฒฐ๋ก  ๋ฐ ํ–ฅํ›„ ์—ฐ๊ตฌ๊ณผ์ œ

1. ์„œ ๋ก 

1.1 ์—ฐ๊ตฌ์˜ ๋ฐฐ๊ฒฝ ๋ฐ ๋ชฉ์ 

์˜ค๋žœ๊ธฐ๊ฐ„๋™์•ˆ ๊ตํ†ต์‚ฌ๊ณ ์˜ ํŠน์„ฑ์„ ํŒŒ์•…ํ•˜๊ธฐ ์œ„ํ•ด ๋งŽ์€ ์—ฐ๊ตฌ๋“ค์ด ์ง„ํ–‰๋˜์–ด์™”์œผ๋ฉฐ, ๋Œ€ํ‘œ์ ์ธ ์—ฐ๊ตฌ๊ฐ€ ๊ตํ†ต์‚ฌ๊ณ ์™€ ๋„๋กœ๊ธฐํ•˜๊ตฌ์กฐ์™€์˜ ๊ด€๊ณ„ํŒŒ์•…์ด๋‹ค. ์ด๋Ÿฌํ•œ ์—ฐ๊ตฌ๋“ค์„ ํ†ตํ•ด์„œ ๊ตํ†ต์‚ฌ๊ณ ์— ์˜ํ–ฅ์„ ๋ฏธ์น˜๋Š” ๋„๋กœ๊ธฐํ•˜๊ตฌ์กฐ ์š”์†Œ๋ฅผ ํŒŒ์•…ํ•˜์—ฌ ์ด๋ฅผ ๊ฐœ์„ ํ•จ์œผ๋กœ์จ ๊ตํ†ต์‚ฌ๊ณ ๋ฅผ ์ค„์ด๋Š” ๋…ธ๋ ฅ์„ ํ•˜์—ฌ์™”๋‹ค. ์ด๋Ÿฌํ•œ ๋…ธ๋ ฅ์€ ์ฃผ๋กœ ํ†ต๊ณ„๋ชจํ˜•์„ ํ†ตํ•ด์„œ ์ด๋ฃจ์–ด์ ธ ์™”์œผ๋ฉฐ, ์ด๋ฅผ ์œ„ํ•ด ์‚ฌ์šฉ๋œ ์ฃผ์š”ํ†ต๊ณ„๋ชจํ˜•์€ ๊ฐ€์‚ฐ๋ชจํ˜•(Count Model)์ด๋ฉฐ, ์„ ํ˜•ํšŒ๊ท€๋ชจํ˜•(Linear Regression Model), ํฌ์•„์†ก(Poisson Model) ๊ทธ๋ฆฌ๊ณ  ์Œ์ดํ•ญ๋ชจํ˜•(Negative Binomial Model)์ด ๋Œ€ํ‘œ์ ์ธ ๊ฐ€์‚ฐ๋ชจํ˜•์ด๋‹ค.

ํ•˜์ง€๋งŒ ์ด๋Ÿฌํ•œ ๋ชจํ˜•๋“ค์€ ๋ชจ๋“  ๊ด€์ธก๊ฐ’(๋„๋กœ๊ตฌ๊ฐ„ ํ˜น์€ ๊ต์ฐจ๋กœ ๋“ฑ)์— ๋Œ€ํ•œ ์ถ”์ •๋œ ๋ชจ์ˆ˜(parameter)๊ฐ€ ๊ณ ์ •๋˜์–ด์žˆ๋‹ค๊ณ  ๊ฐ€์ •ํ•˜๊ณ  ์žˆ๋‹ค. ์ฆ‰, ๋„๋กœ์˜ ๊ตฌ๊ฐ„์— ์„ค์น˜๋˜์–ด ์žˆ๋Š” ๊ธฐํ•˜๊ตฌ์กฐ์˜ ๊ธธ์ดโ€คํญ๋“ฑ์˜ ์„ค์น˜ํ˜•ํƒœ์— ๊ด€๊ณ„์—†์ด ๋™์ผํ•œ ๊ฐ’์ด ์ ์šฉ๋˜์–ด ์‚ฌ์šฉ๋˜์–ด ์™”๋‹ค. ๋„๋กœ๊ธฐํ•˜๊ตฌ์กฐ์˜ ๊ฒฝ์šฐ, ํ˜„์‹ค์ ์œผ๋กœ ๋™์ผํ•œ ๊ธฐํ•˜๊ตฌ์กฐ๊ฐ€ ๋„๋กœ์ „์ฒด์— ์ ์šฉ๋  ์ˆ˜๊ฐ€ ์—†์Œ์—๋„ ๋ถˆ๊ตฌํ•˜๊ณ , ์ง€๊ธˆ๊นŒ์ง€ ๋ชจํ˜•์— ์ ์šฉ๋œ ๋ณ€์ˆ˜๋“ค์€ ์ด๋Ÿฌํ•œ ํ˜„์‹ค์„ ๊ณ ๋ คํ•˜์ง€ ์•Š์€์ฑ„ ์‚ฌ์šฉ๋˜๊ณ  ์žˆ๋Š” ์‹ค์ •์ด๋‹ค. ๊ทธ ๊ฒฐ๊ณผ, ๋ชจํ˜•์„ ํ†ตํ•ด ์ถ”์ •๋œ ๊ณ„์ˆ˜์˜ ํ‘œ์ค€์˜ค์ฐจ๊ฐ’์ด ๊ณผ์†Œ์ถ”์ •๋˜์—ˆ์œผ๋ฉฐ, ์ด๋Š” ๊ณ„์ˆ˜์˜ t-๊ฐ’์„ ๊ณผ๋„ํ•˜๊ฒŒ ๋ถ€ํ’€๋ฆฌ๊ฒŒ ๋˜์–ด ๋ชจํ˜•์˜ ์„ค๋ช…๋ ฅ์ด ๋–จ์–ด์ง€๊ฒŒ ๋˜๋Š” ๊ฒฐ๊ณผ๋ฅผ ์ดˆ๋ž˜ํ•˜๊ฒŒ ๋œ๋‹ค.

๋”ฐ๋ผ์„œ, ์ด๋ฒˆ ์—ฐ๊ตฌ์—์„œ๋Š” ๋„๋กœ์˜ ๊ฐ ๊ตฌ๊ฐ„๋ณ„๋กœ ์ƒ์ดํ•  ์ˆ˜ ์žˆ๋Š” ๊ธฐํ•˜๊ตฌ์กฐ์˜ ์ด์งˆ์„ฑ(heterogeneity)์„ ๊ณ ๋ คํ•˜์—ฌ ๊ธฐํ•˜๊ตฌ์กฐ์™€ ๊ตํ†ต์‚ฌ๊ณ ์™€์˜ ๊ด€๊ณ„๋ฅผ ํŒŒ์•…ํ•˜๊ณ ์ž ํ•œ๋‹ค.

1.2 ์—ฐ๊ตฌ์˜ ๋‚ด์šฉ

์ด ์—ฐ๊ตฌ์—์„œ๋Š” ๊ธฐ์กด์˜ ์Œ์ดํ•ญ๋ชจํ˜•(Fixed Parameter Model, ์ดํ•˜ FPM)๊ณผ ์ด ์—ฐ๊ตฌ์—์„œ ์ œ์‹œํ•˜๊ณ ์ž ํ•˜๋Š” ๋„๋กœ๊ธฐํ•˜๊ตฌ์กฐ์˜ ์ด์งˆ์„ฑ์„ ๊ณ ๋ คํ•œ ์Œ์ดํ•ญ๋ชจํ˜•(Random Parameter Model, ์ดํ•˜ RPM)์˜ ๋น„๊ต๋ฅผ ํ†ตํ•ด ๋ณด๋‹ค ํ˜„์‹ค์ ์ธ ๋ฐฉ๋ฒ•์„ ์ œ์‹œํ•˜๊ณ ์ž ํ•œ๋‹ค. ์ด ๋‘๊ฐ€์ง€ ํ˜•ํƒœ์˜ ๋ชจํ˜•์€ ๋„์ถœ๋œ ๊ณ„์ˆ˜๊ฐ’ ๋ฐ ๊ณ„์ˆ˜๊ฐ’๋“ค์˜ ํƒ„๋ ฅ์„ฑ, ํ•œ๊ณ„ํšจ๊ณผ์™€ ํ•จ๊ป˜ RMSE(Root Mean Square Error : ํ‰๊ท ์ œ๊ณฑ๊ทผ์˜ค์ฐจ), MAPE (Mean Absolute Perchange Error : ํ‰๊ท ๋ฐฑ๋ถ„์œจ์˜ค์ฐจ์˜ ํ‰๊ท ), ๊ทธ๋ฆฌ๊ณ  MAE(Mean Absolute Error : ํ‰๊ท ์ ˆ๋Œ€์˜ค์ฐจ)๋ฐฉ๋ฒ•์— ์˜ํ•ด ๋น„๊ต๋˜์—ˆ๋‹ค. ์ž์„ธํ•œ ์—ฐ๊ตฌ ์ˆ˜ํ–‰์ ˆ์ฐจ๋Š” Fig. 1์— ์ œ์‹œ๋˜์–ด ์žˆ๋‹ค.

PIC9F50.gif

Fig. 1. Research Flow

1.3 ์—ฐ๊ตฌ์˜ ๋ฒ”์œ„

๋ณธ ์—ฐ๊ตฌ์˜ ๊ณต๊ฐ„์ โ€ค์‹œ๊ฐ„์ โ€ค๋‚ด์šฉ์  ๋ฒ”์œ„๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

1.3.1 ๊ณต๊ฐ„์  ๋ฒ”์œ„

๋ฏธ ์›Œ์‹ฑํ„ด ์ฃผ(Washington State)์— ์œ„์น˜ํ•œ 7๊ฐœ์˜ ๊ณ ์†๋„๋กœ(interstate highway) : I-5, 82, 90, 182, 205, 405, ๊ทธ๋ฆฌ๊ณ  705(1,528 mileโ‰’2,458 km)

1.3.2 ์‹œ๊ฐ„์  ๋ฒ”์œ„

9๋…„(1999~2007๋…„) ๋™์•ˆ์˜ ๊ณ ์†๋„๋กœ ๊ตํ†ต์‚ฌ๊ณ , ๊ตํ†ต๋Ÿ‰ ๋ฐ ๊ธฐํ•˜๊ตฌ์กฐ ์ž๋ฃŒ

1.3.3 ๋‚ด์šฉ์  ๋ฒ”์œ„

๋„๋กœ๊ตฌ๊ฐ„์— ์„ค์น˜๋œ ๊ธฐํ•˜๊ตฌ์กฐ์˜ ์ด์งˆ์„ฑ์„ ๊ณ ๋ คํ•œ ์Œ์ดํ•ญ๋ชจํ˜• ๋„์ถœ ๋ฐ ๊ฐ ๋ณ€์ˆ˜๋ณ„ ์‚ฌ๊ณ ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ๋ถ„์„

2. ๊ธฐ์กด ๋ฌธํ—Œ ๊ณ ์ฐฐ

2.1 ๊ตํ†ต์‚ฌ๊ณ  ๋ชจํ˜•

๊ธฐ์กด์˜ ์‚ฌ๊ณ ๋ชจํ˜•์€ ์œ ํ˜•๋ณ„๋กœ ํฌ๊ฒŒ 3๊ฐ€์ง€ ํ˜•ํƒœ-์„ ํ˜•ํšŒ๊ท€์‹, ํฌ์•„์†ก ํšŒ๊ท€์‹, ์Œ์ดํ•ญ ํšŒ๊ท€์‹๋กœ ๋‚˜๋ˆ„์–ด์ง„๋‹ค. ์šฐ์„  ์„ ํ˜•ํšŒ๊ท€์‹์€ ์‚ฌ๊ณ  ๋ฐœ์ƒ์— ์˜ํ–ฅ์„ ๋ฏธ์น˜๋Š” ์ธ์ž๋“ค์„ ๋ถ„์„ํ•˜๋Š” ๊ฐ€์žฅ ๋‹จ์ˆœํ•œ ๊ธฐ๋ฒ•์ด๋ฉฐ, ๋‹ค์Œ๊ณผ ๊ฐ™์ด ๋‚˜ํƒ€๋‚ผ ์ˆ˜ ์žˆ๋‹ค.

yi = ฮฑ + ฮฒxi + ฮตi                     (1)

์—ฌ๊ธฐ์„œ, yi : ๊ตฌ๊ฐ„ i์—์„œ์˜ ์‚ฌ๊ณ ๊ฑด์ˆ˜, ์‚ฌ์ƒ์ž์ˆ˜ ๋˜๋Š” ์‚ฌ๊ณ ์œจ(์ข…์†๋ณ€์ˆ˜)

     ฮฑ์™€ ฮฒ : ์ƒ์ˆ˜์™€ ํšŒ๊ท€๊ณ„์ˆ˜(์ถ”์ •๊ฐ’)

     xi : ๊ตฌ๊ฐ„ i์—์„œ์˜ ์‚ฌ๊ณ ์š”์ธ(๋…๋ฆฝ๋ณ€์ˆ˜)

     ฮตi : ํ™•๋ฅ ์˜ค์ฐจํ•ญ N~(0, ฯƒ2)

์‚ฌ๊ณ ๋ชจํ˜• ์ค‘ ๊ฐ€์žฅ ๋‹จ์ˆœํ•œ ๋ฐฉ๋ฒ•์ด๊ธด ํ•˜์ง€๋งŒ, ๊ณ„์ˆ˜ ์ถ”์ •์— ์‚ฌ์šฉ๋˜๋Š” ๋ณ€์ˆ˜๊ฐ’์ด ์ฆ๊ฐ€ํ• ์ˆ˜๋ก ๋ถ„์‚ฐ๊ฐ’ ๋˜ํ•œ ์ฆ๊ฐ€ํ•˜๊ฒŒ ๋˜์–ด ์„ ํ˜•ํšŒ๊ท€์‹์˜ ๊ธฐ๋ณธ๊ฐ€์ •์ธ ๋™๋ถ„์‚ฐ์„ฑ(Homoscedastisity)์„ ์œ„๋ฐฐํ•˜๊ฒŒ ๋œ๋‹ค. ๊ทธ ๊ฒฐ๊ณผ ๋ณ€์ˆ˜์˜ ์œ ์˜์ˆ˜์ค€์— ๋ณ€ํ™”๋ฅผ ์ฃผ๊ฒŒ ๋˜๋ฉฐ, ์ด๋Š” ๋ณ€์ˆ˜์— ๋Œ€ํ•œ ํ†ต๊ณ„์  ์œ ์˜์„ฑ์„ ๋‚ฎ์ถ”๊ฒŒ ๋œ๋‹ค. ๋ฟ๋งŒ ์•„๋‹ˆ๋ผ ๊ตํ†ต์‚ฌ๊ณ ์ˆ˜์™€ ๊ฐ™์€ ์–‘์˜ ๋ณ€์ˆ˜์— ๋Œ€ํ•ด ์Œ(Negative)์˜ ์‚ฌ๊ณ ์ˆ˜๋ฅผ ์˜ˆ์ธกํ•œ๋‹ค๋Š” ๋‹จ์ ์ด ์žˆ๋‹ค. ํŠนํžˆ ๊ตํ†ต์‚ฌ๊ณ ์˜ ํŠน์„ฑ์ธ ํŠน์ •๊ธฐ๊ฐ„ ๋™์•ˆ ์‚ฌ๊ณ ๊ฐ€ ๋ฐœ์ƒํ•˜์ง€ ์•Š์•˜๊ฑฐ๋‚˜ ๋‚ฎ์€ ์‚ฌ๊ณ ๊ฑด์ˆ˜์— ๋Œ€ํ•ด์„  ์Œ์˜ ๊ฐ’์„ ์˜ˆ์ธกํ•˜๊ฒŒ ๋œ๋‹ค(Jovanis and Chang, 1986). ๋˜ํ•œ, Miaou et al.(1993)์€ ๊ธฐ์กด์˜ ์„ ํ˜•๋ชจํ˜•์€ ์‚ฐ๋ฐœ์ (sporadic)์ด๊ณ  ๋ฌด์ž‘์œ„(random)๋กœ ๋ฐœ์ƒํ•˜๋Š” ์‚ฌ๊ณ ์˜ ๋ชจํ˜•ํ™”์—๋Š” ์ ํ•ฉํ•˜์ง€ ์•Š๋‹ค๊ณ  ์ฃผ์žฅํ•˜์˜€๋‹ค. ์ด๋Ÿฌํ•œ ์Œ์˜ ์‚ฌ๊ณ ์˜ˆ์ธก๊ฐ’์— ๋Œ€ํ•œ ๋ฌธ์ œ๋ฅผ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•œ ํ•œ ๊ฐ€์ง€ ๋ฐฉ๋ฒ•์œผ๋กœ 0์˜ ๊ฐ’์„ ๋‚˜ํƒ€๋‚ด๋Š” ์‚ฌ๊ณ ๋ฅผ ์ œ๊ฑฐํ•˜๊ณ  ๋ถ„์„ํ•˜๋Š” ๊ธฐ๋ฒ•(Left-Truncating the Accident Frequency at Zero)์„ ์‚ฌ์šฉํ•  ์ˆ˜ ์žˆ์œผ๋‚˜, ์ด๋Š” ๋ชจํ˜• ์ถ”์ •์— ์‚ฌ์šฉ๋˜๋Š” ํ‘œ๋ณธ์ˆ˜๊ฐ€ ์ค„์–ด๋“ค ๋ฟ๋งŒ ์•„๋‹ˆ๋ผ, ๋ชจํ˜•์˜ ๊ฐœ๋…์ด ํƒ€ ๋ชจํ˜•์— ๋น„ํ•ด ์—ด์•…ํ•ด์ง„๋‹ค. ์ด๋Ÿฌํ•œ ๋ฌธ์ œ์ ์„ ํ•ด๊ฒฐํ•˜๊ณ ์ž, Jovanis and Chang(1986), ๊ทธ๋ฆฌ๊ณ  Joshua and Garber(1990)๋“ฑ์€ ์‚ฌ๊ณ ๊ฑด์ˆ˜๋ฅผ ์ด์‚ฐ์  ํ™•๋ฅ ๋ณ€์ˆ˜(Discrete Random Variable)๋กœ ํ•ด์„ํ•˜๋Š” ํฌ์•„์†กํšŒ๊ท€์‹(Poisson Regression)์„ ์ œ์•ˆํ•˜์˜€์œผ๋ฉฐ, ํฌ์•„์†กํšŒ๊ท€์‹์˜ ์ผ๋ฐ˜์‹์€ ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

PIC9FAE.gif                    (2)

์—ฌ๊ธฐ์„œ, P(ni) : ์‚ฌ๊ณ  n์ด ์ง€์  i์—์„œ ๋ฐœ์ƒํ•  ํ™•๋ฅ 

      ฮป   : ํ‰๊ท ์‚ฌ๊ณ ๋ฐœ์ƒ๊ฑด์ˆ˜(=exp(ฮฒXi))

      ฮฒ   : ์ถ”์ •๋œ ๊ณ„์ˆ˜

      Xi  : ์ง€์  i์˜ ์†์„ฑ(๊ตํ†ต๋Ÿ‰, ๋„๋กœ ๊ธฐํ•˜๊ตฌ์กฐโ€คํ™˜๊ฒฝ๋“ฑ)

ํฌ์•„์†ก๋ชจํ˜•์€ โ€œ๋ถ„์‚ฐ๊ณผ ํ‰๊ท ์ด ๊ฐ™๋‹ค.โ€๋ผ๋Š” ์ค‘์š”ํ•œ ๊ธฐ๋ณธ์ „์ œ๊ฐ€ ์žˆ๋‹ค. ํ•˜์ง€๋งŒ, ์‹ค์ œ ๋ฐœ์ƒํ•˜๋Š” ๊ตํ†ต์‚ฌ๊ณ ๋Š” ๋Œ€๋ถ€๋ถ„์˜ ๊ฒฝ์šฐ, ๋ถ„์‚ฐ๊ฐ’์ด ํ‰๊ท ๊ฐ’๋ณด๋‹ค ํฐ ๊ณผ๋ถ„์‚ฐ(Overdispersion)์˜ ํ˜•ํƒœ๋ฅผ ๋ณด์ด๊ฒŒ ๋˜์–ด ํฌ์•„์†ก ๋ชจํ˜•์ด ์ ํ•ฉํ•˜์ง€ ์•Š์€ ๊ฒฝ์šฐ๊ฐ€ ๋ฐœ์ƒํ•˜๊ฒŒ ๋œ๋‹ค. ์ด๋ฅผ ๋ณด์™„ํ•˜๊ธฐ ์œ„ํ•ด ์Œ์ดํ•ญํšŒ๊ท€์‹(Negative Binomial Regression Model)์ด ์ฃผ๋กœ ์‚ฌ์šฉ๋˜๊ณ  ์žˆ๋‹ค. ์Œ์ดํ•ญ๋ถ„ํฌ์—๋Š” ์‚ฌ๊ณ ์ˆ˜(ฮปi)์™€ ์˜ค์ฐจํ•ญ(ฮตi)์ด ํฌํ•จ๋˜๋ฉฐ ์ด๋Š” ๋‹ค์Œ์‹๊ณผ ๊ฐ™๋‹ค.

ฮปi = exp(ฮฒXi + ฮตi)                     (3)

์—ฌ๊ธฐ์„œ, exp(ฮตi) : ํ‰๊ท ์ด 1์ด๊ณ  ๋ถ„์‚ฐ์ด ฮฑ์ธ ๊ฐ๋งˆ๋ถ„ํฌ๋ฅผ ๋”ฐ๋ฅด๋Š” ์˜ค์ฐจํ•ญ

์Œ์ดํ•ญ ๋ชจํ˜•์€ ํฌ์•„์†ก๋ชจํ˜•์˜ ํ™•์žฅ๋œ ํ˜•ํƒœ์ด๋ฉฐ, ๋งŒ์•ฝ ์˜ค์ฐจํ•ญ์˜ ฮฑ๊ฐ’์ด ํ†ต๊ณ„์ ์œผ๋กœ 0๊ณผ ๋‹ค๋ฅธ ๊ฒฝ์šฐ์—๋Š” ์Œ์ดํ•ญ ๋ชจํ˜•์ด ์ ํ•ฉํ•˜๋ฉฐ, ๋ฐ˜๋Œ€์˜ ๊ฒฝ์šฐ(ฮฑ=0)์—๋Š” ํฌ์•„์†ก๋ชจํ˜•์ด ์ ํ•ฉํ•˜๊ฒŒ ๋œ๋‹ค. Engel (1984), Lawless (1987), Maher (1991) ๊ทธ๋ฆฌ๊ณ  Miaou (1994)๋Š” ๊ตํ†ต์‚ฌ๊ณ ์™€ ๊ธฐํ•˜๊ตฌ์กฐ์™€์˜ ๊ด€๊ณ„ ํŒŒ์•…์„ ์œ„ํ•œ ์—ฐ๊ตฌ์—์„œ ์Œ์ดํ•ญ๋ชจํ˜•์ด ์„ ํ˜• ๋˜๋Š” ํฌ์•„์†ก ๋ชจํ˜•๋ณด๋‹ค ์šฐ์ˆ˜ํ•˜๋‹ค๋Š” ์—ฐ๊ตฌ๊ฒฐ๊ณผ๋ฅผ ๋ฐœํ‘œํ•˜์˜€๋‹ค.

2.2 ๊ตํ†ต์‚ฌ๊ณ ์™€ ๊ธฐํ•˜๊ตฌ์กฐ์™€์˜ ๊ด€๊ณ„

๊ตํ†ต์‚ฌ๊ณ ์™€ ๊ธฐํ•˜๊ตฌ์กฐ์™€์˜ ๊ด€๊ณ„์— ๊ด€ํ•œ ์—ฐ๊ตฌ๋Š” ๋งŽ์€ ์—ฐ๊ตฌ์ž๋“ค์— ์˜ํ•ด ์ด๋ฃจ์–ด์ ธ ์™”๋Š”๋ฐ, ์—ฐ๊ตฌ์— ์‚ฌ์šฉ๋œ ๊ธฐํ•˜๊ตฌ์กฐ๋Š” ๊ตํ†ต๋Ÿ‰์„ ํฌํ•จํ•˜์—ฌ ๊ธธ์–ด๊นจํญ(Ogden, 1997), ์ฐจ์„ ์ˆ˜์™€ ์ฐจ์„ ๋„“์ด, ๊ทธ๋ฆฌ๊ณ  ์ œํ•œ์†๋„(Noland, 2003), ์ข…โ€คํšก๋‹จ ๊ณก์„  ๊ตฌ์„ฑ์š”์†Œ์˜ ์ตœ์†Œโ€ค์ตœ๋Œ€๊ฐ’(Zhang et al., 2005), ์‹œ๊ฑฐ, ๊ตฌ๋ฐฐ(Caliendo et al., 2007), ์ธํ„ฐ์ฒด์ธ์ง€ ๋žจํ”„(Montella et al., 2008)๋“ฑ์„ ์‚ฌ๊ณ ์— ๋ฏธ์น˜๋Š” ๋ณ€์ˆ˜๋กœ ์ฑ„ํƒํ•˜์—ฌ ์ฃผ๋กœ ์Œ์ดํ•ญ ๋ชจํ˜•์œผ๋กœ ๋ถ„์„์„ ํ•˜์˜€๋‹ค. ํ•˜์ง€๋งŒ, ์ถ”์ •๋œ ๊ณ„์ˆ˜๊ฐ’๋“ค์ด ๋„๋กœ์˜ ๊ตฌ๊ฐ„์— ๊ด€๊ณ„์—†์ด ๋ชจ๋“  ๊ตฌ๊ฐ„์— ๋Œ€ํ•ด ๊ณ ์ •๋œ ๊ฐ’์œผ๋กœ ๋„๋กœ์˜ ๊ฐ ๊ตฌ๊ฐ„์ด ๊ฐ€์งˆ ์ˆ˜ ์žˆ๋Š” ๊ฐ๊ฐ์˜ ์ด์งˆ์ ์ธ ํŠน์„ฑ์— ๊ด€ํ•œ ๊ฐ€๋Šฅ์„ฑ์€ ๋ฐฐ์ œํ•˜์˜€๋‹ค๋Š” ๋‹จ์ ์ด ์žˆ๋‹ค.

์ด๋Ÿฌํ•œ ์ด์งˆ์„ฑ๊ณผ ๊ด€๋ จ๋œ ์—ฐ๊ตฌ์ธก๋ฉด์—์„œ๋Š” Shankar et al.(1998)์€ 2๊ฐ€์ง€ ํ˜•ํƒœ์˜ ๋ชจํ˜•-Random Effects Negative Binomial (RENB)๊ณผ Cross-sectional Negative Binomial์„ ์ด์šฉํ•˜์—ฌ, RENB ๋ชจํ˜•์ด ์ƒ์ˆ˜๊ฐ’์—์„œ ์ด์งˆ์„ฑ์ด ๋„์ถœ๋˜๋„๋ก ํ•˜์˜€์œผ๋ฉฐ, Chin et al.(2003)์€ ์‹ฑ๊ฐ€ํฌ๋ฅด์˜ ๊ต์ฐจ๋กœ์—์„œ ๋ฐœ์ƒํ•œ ์‚ฌ๊ณ ๋ฅผ ์ด์šฉํ•˜์—ฌ, ๊ด€์ธก๋˜์ง€ ์•Š์€ ์ด์งˆ์„ฑ(Unobserved Heterogeneity)๊ณผ ์‹œ๊ณ„์—ด ์ƒ๊ด€(serial correlation)์„ ๋‹ค๋ฃจ๊ธฐ ์œ„ํ•ด RENB๋ชจํ˜•์„ ์ ์šฉํ•˜์˜€๋‹ค. Washington et al.(2010)์€ ๊ณ„์ˆ˜๊ฐ’์ด ๊ด€์ธก๊ฐ’์— ๋”ฐ๋ผ ๋‹ค์–‘ํ•œ ๊ฐ’์„ ๋‚˜ํƒ€๋‚ด์ง€๋งŒ, ๊ณ ์ •๋œ ๊ฐ’์ด๋ผ๊ณ  ๊ฐ€์ •ํ•˜์—ฌ ์ด๋ฅผ ์ œ์•ฝํ•  ๊ฒฝ์šฐ, ์ผ๊ด€์„ฑ์ด ์—†๊ณ (inconsistent) ํŽธ์ค‘๋œ(unbiased) ๊ฒฐ๊ณผ๊ฐ€ ๋„์ถœ๋จ์„ ์ฆ๋ช…ํ•˜์˜€๋‹ค.

3. ์ž๋ฃŒ๊ตฌ์ถ•

3.1 ๋ถ„์„๋ฒ”์œ„

์ด ์—ฐ๊ตฌ์—์„œ ์‚ฌ์šฉ๋œ ๊ตํ†ต์‚ฌ๊ณ , ๊ตํ†ต๋Ÿ‰ ๋ฐ ๊ธฐํ•˜๊ตฌ์กฐ ์ž๋ฃŒ๋Š” ์ด 9๋…„ ๋™์•ˆ์˜ ๋ฏธ๊ตญ ์›Œ์‹ฑํ„ด ์ฃผ(Washington State)์— ์†ํ•œ ๊ณ ์†๋„๋กœ(interstate highway)๋ฅผ ๋Œ€์ƒ์œผ๋กœ ๋ถˆ๊ท ํ˜• ํŒจ๋„๋ฐ์ดํƒ€(Unbalanced Panel Data)1) ํ˜•์‹์œผ๋กœ ๊ตฌ์ถ•๋˜์—ˆ๋‹ค. ๋ฐ์ดํ„ฐ ๊ตฌ์ถ•์— ํฌํ•จ๋œ ๊ณ ์†๋„๋กœ๋Š” I-5, I-82, I-90, I-182, I-205, I-405, ๊ทธ๋ฆฌ๊ณ  I-705์˜ 7๊ฐœ ๋…ธ์„ ์ด๋ฉฐ, ๋ณด๋‹ค ์ž์„ธํ•œ ๋…ธ์„ ์˜ ๊ธธ์ด ๋ฐ ์œ„์น˜๋Š” Fig. 2์— ๋‚˜ํƒ€๋‚˜์žˆ๋‹ค.

๋ชจ๋“  ๊ณ ์†๋„๋กœ๋Š” ๋ฐฉํ–ฅ๋ณ„๋กœ ๋ณธ์„ ๊ตฌ๊ฐ„๊ณผ ์ธํ„ฐ์ฒด์ธ์ง€ ๊ตฌ๊ฐ„์œผ๋กœ ๊ตฌ๋ถ„๋˜์–ด์žˆ๋Š”๋ฐ, ์ด๋Š” ๊ธฐ๋ณธ์ ์œผ๋กœ ๋‘ ๊ตฌ๊ฐ„์˜ ๊ธฐํ•˜๊ตฌ์กฐ๊ฐ€ ์ƒ์ดํ•˜๊ธฐ ๋•Œ๋ฌธ์ž„๊ณผ ๋™์‹œ์— ๊ตํ†ต์‚ฌ๊ณ  ๋ถ„์„์—์„œ ์ฃผ์š”ํ•œ ์š”์ธ์ธ ๊ตํ†ต๋Ÿ‰์ด ์ธํ„ฐ์ฒด์ธ์ง€๋ฅผ ๊ธฐ์ค€์œผ๋กœ ๋ฐ”๋€Œ๊ธฐ ๋•Œ๋ฌธ์ด๋‹ค. ๊ตฌ๊ฐ„ ๋ถ„ํ• ์˜ ๊ฒฝ์šฐ, ๋˜ ๋‹ค๋ฅธ ํ˜•ํƒœ๋กœ ๊ฐ€๋Šฅ(๊ธฐํ•˜๊ตฌ์กฐ๊ฐ€ ๋ณ€๊ฒฝ ๋˜๋Š” ์ง€์  ๋“ฑ)ํ•˜์ง€๋งŒ, ์ด ์—ฐ๊ตฌ์—์„œ๋Š” ์ด์งˆ์„ฑ์„ ๊ณ ๋ คํ•˜์˜€๊ธฐ ๋•Œ๋ฌธ์— ๋ชจํ˜•์—์„œ ๋„์ถœ๋˜๋Š” ๊ณ„์ˆ˜๊ฐ’ ๋“ฑ์„ ํ†ตํ•ด ์„ค๋ช…์ด ๋˜์–ด์ง€๋ฉฐ, Milton ๋“ฑ(2008)์˜ ์—ฐ๊ตฌ์—์„œ๋„ ๊ธฐ์ˆ ๋˜์–ด ์žˆ๋‹ค.

์ธํ„ฐ์ฒด์ธ์ง€ ๊ตฌ๊ฐ„์€ ๋ณธ์„ ์—์„œ ์ธํ„ฐ์ฒด์ธ์ง€๋กœ์˜ ๋ถ„๋ฅ˜๋ถ€ ์‹œ์ ์—์„œ ์ธํ„ฐ์ฒด์ธ์ง€์—์„œ ๋ณธ์„ ์œผ๋กœ์˜ ํ•ฉ๋ฅ˜๋ถ€ ์‹œ์ ๊นŒ์ง€์˜ ๊ตฌ๊ฐ„์ด๋ฉฐ, ๋ณธ์„ ๊ตฌ๊ฐ„์€ ๊ทธ ์™ธ์˜ ๊ตฌ๊ฐ„์œผ๋กœ ์ •์˜ํ•˜์˜€๋‹ค. ์ด๋Ÿฌํ•œ ์ „์ œ์กฐ๊ฑด ํ•˜์— ์ด 7๊ฐœ๋…ธ์„ ์˜ 1,528mile(2,458km)๊ตฌ๊ฐ„์ด 1,168๊ฐœ์˜ ๊ตฌ๊ฐ„์œผ๋กœ ๊ตฌ๋ถ„๋˜์—ˆ์œผ๋ฉฐ, ๊ตฌ๊ฐ„์˜ ๋…ธ์„ ๋ณ„ ๋ฐฉํ–ฅ๋ณ„ ๊ฐœ์ˆ˜ ๋ฐ ๊ธธ์ด๋Š” Table 1์— ๋‚˜ํƒ€๋‚˜์žˆ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ ์‚ฌ์šฉ๋œ ๊ธฐํ•˜๊ตฌ์กฐ์— ๋Œ€ํ•œ ์ž๋ฃŒ๋Š” ์ •์˜๋œ ๊ฐ๊ฐ์˜ ๊ตฌ๊ฐ„์— ๋Œ€ํ•œ ๊ฐ’์œผ๋กœ ๋ณ€ํ™˜๋˜์–ด ์ ์šฉ๋˜์—ˆ๋‹ค. ๊ทธ๋ฆฌ๊ณ  ๊ตํ†ต๋Ÿ‰๊ณผ ์‚ฌ๊ณ ๋Š” 9๋…„๋™์•ˆ ๊ฐ ๋…„๋„๋ณ„๋กœ ์ ์šฉ๋˜์—ˆ๋‹ค.

3.2 ๊ฐœ๋ณ„์ž๋ฃŒ๋ถ„์„

Table 2๋Š” ์ด ์—ฐ๊ตฌ์—์„œ ์‚ฌ์šฉ๋œ ํ™•๋ฅ ์  ๋ชจ์ˆ˜๋ฅผ ๊ณ ๋ คํ•œ ์Œ์ดํ•ญ ๋ชจํ˜•์— ์‚ฌ์šฉ๋œ ๋ณ€์ˆ˜์— ๋Œ€ํ•œ ํ†ต๊ณ„๊ฐ’์„ ๋‚˜ํƒ€๋‚ด๊ณ  ์žˆ๋‹ค. ์ฃผ์š”ํ•œ ๋ณ€์ˆ˜๋กœ๋Š” ๋…„๊ฐ„ ์‚ฌ๊ณ ๊ฑด์ˆ˜, ์ฐจ์„ ์ˆ˜, ์ขŒโ€ค์šฐ ๊ธธ์–ด๊นจํญ, ์ข…โ€คํšก๋‹จ ๊ณก์„ ์— ๊ด€ํ•œ ๋ณ€์ˆ˜, ๊ตฌ๊ฐ„์˜ ๊ธธ์ด ๋ฐ ๊ตํ†ต๋Ÿ‰์ด ์‚ฌ์šฉ๋˜์—ˆ์œผ๋ฉฐ, ๋…„๊ฐ„ ์‚ฌ๊ณ ๊ฑด์ˆ˜์˜ ํ‰๊ท ๊ฐ’์€ 10.70 ํ‘œ์ค€ํŽธ์ฐจ๋Š” 18.92 ๊ทธ๋ฆฌ๊ณ  ์ตœ๋Œ€๊ฐ’์€ 388๊ฑด์„ ๋ณด์—ฌ์ฃผ๊ณ  ์žˆ๋‹ค.

์ฐจ์„ ์ˆ˜ ๋ฐ ์ขŒโ€ค์šฐ ๊ธธ์–ด๊นจํญ์˜ ๊ฒฝ์šฐ, ๊ตฌ๊ฐ„๋‚ด์—์„œ ๋™์ผํ•œ ์„ค๊ณ„๊ฐ€ ์œ ์ง€๋˜๋Š” ๊ณณ๋„ ์žˆ๋Š” ๋ฐ˜๋ฉด, ๋‹ฌ๋ผ์ง€๋Š” ๊ฒฝ์šฐ๋„ ๊ณ ๋ คํ•˜๊ธฐ ์œ„ํ•ด ์ด ์—ฐ๊ตฌ์—์„œ๋Š” ๊ตฌ๊ฐ„๋ณ„ ๊ธธ์ด๋น„์œจ๋กœ ์ •์˜ํ•˜์˜€๋‹ค. ๋”ฐ๋ผ์„œ ๊ธฐํ•˜๊ตฌ์กฐ๊ฐ€ ๋™์ผํ•˜๊ฒŒ ์œ ์ง€๋˜๋Š” ๊ตฌ๊ฐ„์—์„œ์˜ ์ตœ๋Œ€๊ฐ’์€ 1์„ ๋„˜์ง€ ์•Š์œผ๋ฉฐ, ์ƒ์ดํ•œ ๊ธฐํ•˜๊ตฌ์กฐ๊ฐ€ ๊ตฌ๊ฐ„๋‚ด์— ๋‹ค์†Œ ์กด์žฌํ•˜๋”๋ผ๋„ ๊ตฌ๊ฐ„๋‚ด์—์„œ์˜ ํ•ฉ์€ ์ตœ๋Œ€๊ฐ’์ธ 1์„ ๋„˜์ง€ ์•Š๊ฒŒ๋œ๋‹ค. ๊ณ ์†๋„๋กœ๋ฅผ ๋Œ€์ƒ์œผ๋กœ ๋ถ„์„์ด ์ง„ํ–‰๋˜์—ˆ๊ธฐ ๋•Œ๋ฌธ์— 2์ฐจ์„  ๋„๋กœ๋Š” ๋งŽ์ด ์กด์žฌํ•˜์ง€ ์•Š์œผ๋ฉฐ, ๊ธธ์ด๋Š” 0.52์˜ ํ‰๊ท ๊ฐ’๊ณผ 0.494์˜ ํ‘œ์ค€ํŽธ์ฐจ๊ฐ’์„ ๊ฐ€์ง„๋‹ค. 3์ฐจ์„  ๋„๋กœ์˜ ๊ฒฝ์šฐ์—๋Š” 0.336, 0.461, 4์ฐจ์„ ์€ 0.135, 0.331 ๊ทธ๋ฆฌ๊ณ  5์ฐจ์„  ๋„๋กœ์˜ ๊ฒฝ์šฐ์—๋Š” 0.002, 0.043์˜ ํ‰๊ท ๊ฐ’๊ณผ ํ‘œ์ค€ํŽธ์ฐจ๊ฐ’์„ ๊ฐ€์ง„๋‹ค.

PICA0E8.gif

Fig. 2. Interstate Highway in Washington State

Table 1. Number of Segments  and Miles by Route and Direction

Route

Increasing  Direction

Interchange Segment

Non-interchange Segment

Miles(Kilometers)

Segments

Miles(Kilometers)

Segments

5

104.84(168.69)

138

171.78(276.39)

139

82

27.51(44.26)

33

105.06(169.04)

34

90

65.73(105.76)

83

231.79(372.95)

83

182

7.08(11.39)

8

8.11(13.05)

9

205

4.14(6.66)

5

6.43(10.35)

6

405

13.53(21.77)

19

16.77(26.98)

20

705

0.5(0.80)

1

1(1.61)

2

Total 

223.33(375.43)

287

540.94(870.37)

293

Route

Decreasing  Direction

Interchange Segment

Non-interchange Segment

Miles(Kilometers)

Segments

Miles(Kilometers)

Segments

5

104.48(168.11)

138

172.13(276.96)

140

82

26.5(42.64)

33

106.07(170.67)

34

90

65.28(105.04)

85

232.24(373.67)

85

182

7.55(12.15)

8

7.64(12.29)

9

205

4.06(6.54)

5

6.51(10.47)

6

405

13.84(22.27)

20

16.46(26.48)

21

705

0.55(0.88)

2

0.95(1.53)

2

Total 

222.26(357.62)

291

542.00(872.08)

297

๊ธธ์–ด๊นจ์˜ ๊ฒฝ์šฐ์—๋Š” 5๊ฐ€์ง€์˜ ๊ฒฝ์šฐ(2-feet ์ดํ•˜, 3~4, 5~9, 10, 10-feet ์ด์ƒ)๋กœ ๊ตฌ๋ถ„ํ•˜์˜€์œผ๋ฉฐ, ์ฐจ์„ ์ˆ˜์™€ ๋งˆ์ฐฌ๊ฐ€์ง€๋กœ ๊ตฌ๊ฐ„๋‚ด์—์„œ ์ฐจ์ง€ํ•˜๋Š” ๋น„์œจ๋กœ ์ •์˜ํ•˜์˜€๋‹ค.

๊ตฌ๊ฐ„๋‚ด์—์„œ ํšก๋‹จ๊ณก์„ ์˜ ๊ฐœ์ˆ˜๋Š” ํ‰๊ท ์ ์œผ๋กœ 1.805๊ฐœ๊ฐ€ ์กด์žฌํ•˜๋Š” ๊ฒƒ์œผ๋กœ ํŒŒ์•…๋˜์—ˆ์œผ๋ฉฐ, ์ตœ๋Œ€ 37๊ฐœ์˜ ํšก๋‹จ๊ณก์„ ์ด ์กด์žฌํ•˜๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค. ๋˜ํ•œ, ๊ฐ€์žฅ ์งง์€ ํšก๋‹จ๊ณก์„ ์€ ํ‰๊ท ์ ์œผ๋กœ 0.182mile (0.29km)์˜ ๊ธธ์ด๋ฅผ ๊ฐ€์ง€๊ณ  ์žˆ์œผ๋ฉฐ, ๋ฐ˜๋ฉด ๊ฐ€์žฅ ๊ธด ํšก๋‹จ๊ณก์„ ์€ 0.275mile(0.44km)์˜ ํ‰๊ท ๊ฐ’์„ ๊ฐ€์ง€๋Š” ๊ฒƒ์œผ๋กœ ํŒŒ์•…๋˜์—ˆ๋‹ค. ํšก๋‹จ๊ณก์„ ์˜ ์ตœ์†Œ๊ต๊ฐ์˜ ํ‰๊ท ๊ฐ’์€ 12.186ยฐ์ด๋ฉฐ, ์ตœ๋Œ€๊ต๊ฐ์€ 21.080ยฐ์˜ ํ‰๊ท ๊ฐ’์„ ๊ฐ€์ง€๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค.

์ข…๋‹จ๊ณก์„ ์˜ ๊ฒฝ์šฐ์—๋Š” 3.125๊ฐœ์˜ ๊ณก์„ ์ด ๊ตฌ๊ฐ„๋‚ด์—์„œ ํ‰๊ท ์ ์œผ๋กœ ์กด์žฌํ•˜๋ฉฐ, ์ตœ๋Œ€๊ฐœ์ˆ˜๋Š” 30๊ฐœ์ธ ๊ฒƒ์œผ๋กœ ํŒŒ์•…๋˜์—ˆ๋‹ค. ์ข…๋‹จ๊ณก์„ ๊ณผ ๊ด€๋ จ๋œ ๋ณ€์ˆ˜๋กœ๋Š” ์ตœ์†Œโ€ค์ตœ๋Œ€ ์ข…๋‹จ๊ตฌ๋ฐฐ๊ฐ€ ์žˆ์œผ๋ฉฐ, ์ตœ์†Œ ๊ตฌ๋ฐฐ๋Š” 1.232%, 1.385%์˜ ํ‰๊ท ๊ฐ’๊ณผ ํ‘œ์ค€ํŽธ์ฐจ๊ฐ’์„ ๊ฐ€์ง€๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚œ ๋ฐ˜๋ฉด, ์ตœ๋Œ€ ๊ตฌ๋ฐฐ์˜ ๊ฒฝ์šฐ์—๋Š” 2.78%์˜ ํ‰๊ท ๊ฐ’๊ณผ 2.075%์˜ ํ‘œ์ค€ํŽธ์ฐจ๊ฐ’์„ ๊ฐ€์ง€๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค.

๋Œ€์ƒ๊ตฌ๊ฐ„์˜ ํ‰๊ท  ๊ธธ์ด๋Š” 1.309mile(2.11km)์ด๋ฉฐ, ๊ตฌ๊ฐ„๋‚ด ์—ฐํ‰๊ท ์ผ ๊ตํ†ต๋Ÿ‰(๋ฐฉํ–ฅ๋ณ„ ์ฐจ์„ ๋ณ„)์€ ํ‰๊ท  13,043๋Œ€๋กœ ๋‚˜ํƒ€๋‚ฌ์œผ๋ฉฐ, ์ตœ๋Œ€๊ตํ†ต๋Ÿ‰์€ 44,223๋Œ€๋กœ ์กฐ์‚ฌ๋˜์—ˆ๋‹ค. ๊ตฌ๊ฐ„์˜ ๊ธธ์ด์™€ ๊ตํ†ต๋Ÿ‰์€ ๋กœ๊ทธ๊ฐ’์œผ๋กœ ๋ณ€ํ™˜๋œ ๊ฐ’์ด ๋ชจํ˜•์— ์ ์šฉ๋˜์—ˆ๋‹ค.

4. ๋ชจํ˜•์˜ ๊ตฌ์ถ• ๋ฐ ๊ฒฐ๊ณผ๋ถ„์„

Table 2. Statistics of Accident, Geometrics and ADT Variables

Variable

Variable Description

Mean

Std.Dev.

Minimum

Maximum

ACC

Number of Accidents per year

10.703

18.917

0

388

NLN2

Portion of segment with 2 lanes

0.525

0.494

0

1

NLN3

Portion of segment with 3 lanes

0.336

0.461

0

1

NLN4

Portion of segment with 4 lanes

0.135

0.331

0

1

NLN5

Portion of segment with 5 lanes

0.002

0.043

0

1

LSHW2

2-ft(60cm) left-shoulder width segment proportion

0.175

0.337

0

1

LSHW34

3- to 4-ft(90~120cm) left-shoulder width segment proportion

0.199

0.387

0

1

LSHW59

5- to 9-ft(150~270cm) left-shoulder width segment proportion

0.133

0.320

0

1

LSHW10

10-ft(300cm) left-shoulder width segment proportion

0.472

0.478

0

1

LSHW1126

Over 10-ft(over 300cm) left-shoulder width segment proportion

0.020

0.132

0

1

RSHW2

2-ft(60cm) right-shoulder width segment proportion

0.178

0.341

0

1

RSHW34

3- to 4-f(90~120cm)t right-shoulder width segment proportion

0.207

0.392

0

1

RSHW59

5- to 9-f(150~270cm)t right-shoulder width segment proportion

0.121

0.308

0

1

RSHW10

10-ft(300cm) right-shoulder width segment proportion

0.464

0.477

0

1

RSHW1124

Over 10-ft(over 300cm) right-shoulder width segment proportion

0.030

0.155

0

1

NHORZ

Number of hirizontal curves in segment

1.807

2.459

0

37

MINHLMI

Shortest horizontal curve-in-segment length (miles)

0.182(0.29km)

0.193

0

1.219

MAXHLMI

Longest horizontal curve-in-segment length (miles)

0.275(0.44km)

0.277

0

2.402

MINHDEL

Smallest horizontal curve-in-segment central angle (degrees)

12.186

14.692

0

98.426

MAXHDEL

Largest horizontal curve-in-segment central angle (degrees)

21.080

21.001

0

111.294

NVERT

Number of vertical curves in segment

3.125

3.210

0

30

MINVCRVG

Smallest absolute vertical curve gradient (%)

1.232

1.385

0

7.430

MAXVCRVG

Largest absolute vertical curve gradient (%)

2.780

2.075

0

10

LENGTH

Segment length (miles)

1.309(2.11km)

1.753

0.01

20.380

AADT

Annual Average Daily Traffic Volume

13043.300

8378.490

916.45

44223.800

4.1 ํ™•๋ฅ ์  ๋ชจ์ˆ˜๋ฅผ ๊ณ ๋ คํ•œ ์Œ์ดํ•ญ๋ชจํ˜• (Random Parameter Negative Binomial)

์ด ์—ฐ๊ตฌ์—์„œ ์ œ์‹œํ•˜๊ณ ์ž ํ•˜๋Š” ํ™•๋ฅ ์  ๋ชจ์ˆ˜๋ฅผ ๊ณ ๋ คํ•œ ์Œ์ดํ•ญ ๋ชจํ˜•์ด ๊ธฐ์กด์˜ ์Œ์ดํ•ญ ๋ชจํ˜•๊ณผ ๋‹ค๋ฅธ ๊ฐ€์žฅ ํฐ ํŠน์ง•์€ ๊ณ„์ˆ˜๊ฐ’(ฮฒ)์ด ๊ณ ์ •๋˜์–ด ์žˆ์ง€ ์•Š๋‹ค๋Š” ๊ฒƒ์ด๋‹ค. ์ผ๋ฐ˜์ ์œผ๋กœ ๋…๋ฆฝ๋ณ€์ˆ˜(x)๋Š” ๊ด€์ธก๋˜๊ฑฐ๋‚˜ ์ˆ˜์ง‘๋œ ์ž๋ฃŒ์— ์ œ์•ฝ๋˜์–ด ์žˆ๊ธฐ ๋•Œ๋ฌธ์— ๋…๋ฆฝ๋ณ€์ˆ˜๋กœ๋ถ€ํ„ฐ ์ด์šฉ๊ฐ€๋Šฅํ•œ ์ตœ๋Œ€ํ•œ์˜ ์ •๋ณด๋ฅผ ์–ป๋Š”๊ฒƒ์— ๋ชจ๋ธ๋ง์˜ ๋ชฉ์ ์ด ์žˆ์œผ๋ฉฐ, ์ด๋Ÿฌํ•œ ๊ด€์ ์—์„œ ๊ฐ ๊ณ„์ˆ˜๋ฅผ ๊ณ ์ •๋œ ๊ฐ’์ด ์•„๋‹ˆ๋ผ, ๊ณต๊ฐ„๋ณ„(i) ํ˜น์€ ์‹œ๊ฐ„๋ณ„(t)๋กœ ๋‹ค์–‘ํ•˜๋‹ค๊ณ  ๊ฐ€์ •ํ•˜์—ฌ ์ ‘๊ทผํ•œ๋‹ค๋ฉด ์ด์งˆ์„ฑ์„ ๊ณ ๋ คํ•œ ๊ณ„์ˆ˜๊ฐ’์„ ๋„์ถœํ•  ์ˆ˜ ์žˆ๊ฒŒ ๋œ๋‹ค. ์˜ˆ๋ฅผ ๋“ค๋ฉด, ์–ด๋Š ํŠน์ •๋ถ„ํฌ๊ฐ€ ํ™•๋ฅ ๋ชจ์ˆ˜์˜ ๊ณ„์†์ ์ธ ๋ณ€ํ™”๋ฅผ ๊ฐ€์žฅ ์ž˜๋‚˜ํƒ€๋‚ธ๋‹ค๋ผ๊ณ  ๊ฐ€์ •ํ•˜๋Š” ๊ฒƒ์ด๋ฉฐ, ์ด๋Ÿฌํ•œ ๊ฐ€์ •์— ๊ธฐ๋ฐ˜ํ•œ ์šฐ๋„ํ•จ์ˆ˜(likelihood function)๋Š” ๊ณ ์ •๋œ ๊ณ„์ˆ˜๊ฐ€ ์•„๋‹Œ ๋ณ€ํ™”ํ•˜๋Š” ๊ณ„์ˆ˜๋กœ๋ถ€ํ„ฐ ๊ณ„์‚ฐ๋œ ํ™•๋ฅ ์„ ๊ธฐ๋ฐ˜์œผ๋กœ ํ•ด์„œ ๋‹ค์Œ๊ณผ ๊ฐ™์ด ์ˆ˜์ •๋ ์ˆ˜ ์žˆ๊ฒŒ ๋œ๋‹ค.

ฮฒit = ฮฒ + โ–ณhi + ฮ“wit     (4)

์—ฌ๊ธฐ์„œ, h๊ฐ€ ์†ํ•œ ๋‘ ๋ฒˆ์งธ ํ•ญ์€ hi์— ์†ํ•œ ๋ณ€์ˆ˜๋“ค์˜ ํ‰๊ท ๊ฐ’์— ๋Œ€ํ•œ ์ด์งˆ์„ฑ์„ ๋‚˜ํƒ€๋‚ด๋ฉฐ, ์„ธ ๋ฒˆ์งธ ํ•ญ์€ ํ‰๊ท ์œผ๋กœ ๋ถ€ํ„ฐ์˜ ํ™•๋ฅ ํŽธ์ฐจ(random deviation)๋ฅผ ๋‚˜ํƒ€๋‚ด๊ฒŒ ๋œ๋‹ค. ๋”ฐ๋ผ์„œ, Eq. (4)๋Š” ํ™•๋ฅ ์  ์ด์งˆ์„ฑ(parameter heterogeniety)์„ ๋‚˜ํƒ€๋‚ด๋Š” ๊ธฐ๋ณธ์ ์ธ ์‹์ด๋ฉฐ, ํŠนํžˆ ๋ธํƒ€(โ–ณ)๋Š” ๊ณ„์ˆ˜(ฮฒ)์— ์˜ํ–ฅ์„ ๋ฏธ์น˜๋Š” ์™ธ์ƒ๋ณ€์ˆ˜(exogenous variable)๋ฅผ ํฌํ•จํ•˜๋ฉฐ, ๋งŒ์•ฝ ๋ณ€์ˆ˜ ํ‰๊ท ๊ฐ’๋‚ด์— ์ด์งˆ์„ฑ์ด ์™ธ์ƒ๋ณ€์ˆ˜๋กœ ์กด์žฌํ•˜์ง€ ์•Š๋Š”๋‹ค๋ฉด, ์ด ํ•ญ์€ ์‚ฌ๋ผ์ง€๊ฒŒ ๋  ๊ฒƒ์ด๋‹ค. ์ฆ‰, ๋ชจ๋“  ์ด์งˆ์„ฑ์€ ์„ธ ๋ฒˆ์งธ ํ•ญ์— ์˜ํ•ด์„œ ๋ชจํ˜•ํ™” ๋˜์–ด์ง€๊ฒŒ ๋œ๋‹ค. ์ด๋Š” ์‚ฌ์šฉ๋œ ๋…๋ฆฝ๋ณ€์ˆ˜์˜ ํŠน์„ฑ์— ๋”ฐ๋ฅธ ๋ถ„ํฌ์˜ ํ˜•ํƒœ๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์ ์šฉ๋  ๊ฒƒ์ด๋ฉฐ, ๋งŒ์•ฝ ๋…๋ฆฝ๋ณ€์ˆ˜์˜ ํ˜•ํƒœ๊ฐ€ ์ดํ•ญ๋ณ€์ˆ˜(dummy variable)๋ผ๋ฉด, ๊ท ์ผ(uniform)๋ถ„ํฌ๊ฐ€, ์—ฐ์†๋ณ€์ˆ˜(continuous variable)๋ผ๋ฉด ์ •๊ทœ(normal) ๋˜๋Š” ๋กœ๊ทธ์ •๊ทœ(lognormal)๋ถ„ํฌ๊ฐ€ ๋  ๊ฒƒ์ด๋‹ค. ๋”ฐ๋ผ์„œ, ์„ธ ๋ฒˆ์งธ ํ•ญ์ด ํ†ต๊ณ„์ ์œผ๋กœ ์œ ์˜ํ•˜๋ฉด, ๊ณ„์ˆ˜๊ฐ’์ด ํ‘œ์ค€ํŽธ์ฐจ๊ฐ’๊ณผ ํ•จ๊ป˜ ๊ฐ ๊ตฌ๊ฐ„๋ณ„๋กœ ์ƒ์ดํ•œ ๊ฐ’์ด(์ด์งˆ์„ฑ์„ ๊ฐ€์ง„๋‹ค) ์ถ”์ •๋˜๋ฉฐ, ํ†ต๊ณ„์ ์œผ๋กœ ์œ ์˜ํ•˜์ง€ ์•Š์œผ๋ฉด, ์ผ๋ฐ˜์ ์ธ ์Œ์ดํ•ญ๋ชจํ˜•์ด ์ ์šฉ๋˜์–ด ๊ณ„์ˆ˜๊ฐ’์ด ๊ตฌ๊ฐ„์— ์ƒ๊ด€์—†์ด ๊ณ ์ •๋œ ๊ฐ’์œผ๋กœ ์ถ”์ •๋œ๋‹ค.

Table 3. Modeling Estimation Results for Fixed-and Random-Parameters Negative Binomial Models

Variable

Fixed Parameter Model (FPM)

Random Parameter Model (RPM)

Coefficient

t-value

Coefficient

t-value

Exposure and Context

Constant

-8.409

-73.556

-7.624

-73.294

Logarithm of Length of segment in miles

0.743

68.992

0.787

83.210

Standard deviation of parameter distribution

-

0.087

11.882

Logarithm of ADT

1.143

99.726

1.036

98.715

Standard deviation of parameter distribution

-

0.008

4.283

Interchange indicator(1 if segment is an interchange segment; 0 otherwise)

-0.027

-1.991

-0.051

-4.418

Number of Lanes by Length Proportion

Three-lane cross-section segment proportion

0.375

15.947

0.473

23.879

Standard deviation of parameter distribution

-

0.094

7.784

Four-lane cross-section segment proportion

0.902

33.995

0.856

38.592

Standard deviation of parameter distribution

-

0.167

10.530

Five-lane cross-section segment proportion

0.785

3.456

1.190

7.868

Left Shoulder Width by Length Proportion

3- to 4-ft(90~120cm) left-shoulder-width proportion

-0.273

-8.528

-0.231

-8.430

Standard deviation of parameter distribution

-

0.287

16.698

5- to 9-ft(150~270cm) left-shoulder-width proportion

-0.496

-18.081

-0.393

-18.274

10-ft(300cm) left-shoulder-width proportion

-0.416

-20.219

-0.300

-16.131

Standard deviation of parameter distribution

-

0.154

14.000

Right Shoulder Width by Length Proportion

3- to 4-ft(90~120cm) right-shoulder-width proportion

-0.286

-8.347

-0.279

-10.380

5- to 9-ft(150~270cm) right-shoulder-width proportion

-0.319

-13.737

-0.344

-14.733

Standard deviation of parameter distribution

-

0.127

6.930

10-ft(300cm) right-shoulder-width proportion

-0.395

-17.875

-0.320

-18.813

Horizontal-Vertical Curvature

Number of horizontal curves in segment

0.029

7.080

0.019

6.150

Shortest horizontal curve-in-segment length in miles

-0.625

-12.484

-0.514

-12.109

Largest degree of curvature in segment

0.006

13.850

0.005

13.023

Largest vertical curve gradient in segment

0.020

5.327

0.029

8.955

Standard deviation of parameter distribution

-

0.021

13.638

Dispersion parameter for negative binomial distribution

Dispersion parameter

0.348

51.092

5.960

41.550

Number of observations

10512

Log-likelihood with constant only

-105,377.5

Log-likelihood at convergence

-29,210.79

-28,315.69

ฯ2

0.722

0.731

4.2 ๊ฒฐ๊ณผ๋ถ„์„

์ด ์—ฐ๊ตฌ์—์„œ๋Š” Table 2์—์„œ ์ œ์‹œ๋œ ๋ณ€์ˆ˜๋“ค์„ ํ™•๋ฅ ์  ๋ชจ์ˆ˜๋ฅผ ์ด์šฉํ•œ ์Œ์ดํ•ญ๋ชจํ˜•์— ์ด์šฉํ•˜์˜€๋‹ค. ์ข…์†๋ณ€์ˆ˜๋Š” ์›Œ์‹ฑํ„ด ์ฃผ 7๊ฐœ์˜ ๊ณ ์†๋„๋กœ์—์„œ ๋ฐœ์ƒํ•œ ๊ตํ†ต์‚ฌ๊ณ ๊ฑด์ˆ˜๋ฅผ ๋…๋ฆฝ๋ณ€์ˆ˜๋กœ๋Š” ๊ธฐํ•˜๊ตฌ์กฐ, ๊ตํ†ต๋Ÿ‰ ๋ฐ ๊ตฌ๊ฐ„์˜ ๊ธธ์ด๋กœ ์„ค์ •ํ•˜์˜€์œผ๋ฉฐ, ํ†ต๊ณ„ํ”„๋กœ๊ทธ๋žจ์ธ LIMDEP์„ ์ด์šฉํ•˜์—ฌ ๊ธฐ์กด์—ฐ๊ตฌ์—์„œ ์ฃผ๋กœ ์‚ฌ์šฉ๋˜๋Š” ์Œ์ดํ•ญ ๋ชจํ˜•(FPM)๊ณผ ๋ณธ ์—ฐ๊ตฌ์—์„œ ์ œ์‹œํ•˜๊ณ ์ž ํ•˜๋Š” ํ™•๋ฅ ์  ๋ชจ์ˆ˜๋ฅผ ์ด์šฉํ•œ ์Œ์ดํ•ญ ๋ชจํ˜•(RPM)์„ ๋„์ถœํ•˜์˜€๋‹ค. ํฌ์•„์†ก ๋ชจํ˜•๊ณผ ์Œ์ดํ•ญ ๋ชจํ˜•์˜ ์„ ํƒ์— ์žˆ์–ด์„œ ์•ž์„œ ์„ค๋ช…ํ•œ๋ฐ๋กœ ๊ณผ๋ถ„์‚ฐ๊ฐ’(ฮฑ)์ด 0.348, t-๊ฐ’์ด 51.092์ž„์„ ๋ณด์—ฌ ํ†ต๊ณ„์ ์œผ๋กœ 0๊ณผ ๋‹ค๋ฅด๋‹ค๋Š” ์œ ์˜์„ฑ์„ ๋ณด์ด๊ธฐ ๋•Œ๋ฌธ์— ๋„์ถœ๋œ ๋ชจํ˜•์€ ํฌ์•„์†ก๋ถ„ํฌ๋ณด๋‹ค ์Œ์ดํ•ญ๋ถ„ํฌ๋ฅผ ๋”ฐ๋ฅด๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚˜ ๊ฒฐ๊ณผ์ œ์‹œ๋Š” ์Œ์ดํ•ญ๋ชจํ˜•์— ์˜ํ•œ ๊ฒฐ๊ณผ๋งŒ ์ œ์‹œํ•˜๊ณ ์ž ํ•œ๋‹ค.

๋ชจํ˜•์˜ ๋„์ถœ์€ Halton draws๋ฅผ ์ด์šฉํ•œ ์ตœ๋Œ€์šฐ๋„๋ฒ•์„ ์‚ฌ์šฉํ•˜์˜€์œผ๋ฉฐ, Bhat (2003) ๊ทธ๋ฆฌ๊ณ  Milton et al. (2008)์˜ ๋…ผ๋ฌธ์—์„œ ์ •ํ™•ํ•œ ๊ณ„์ˆ˜์ถ”์ •์„ ํ•  ์ˆ˜ ์žˆ๋‹ค๊ณ  ์–ธ๊ธ‰๋œ 200 Halton Draws๋ฅผ ์ ์šฉํ•˜์˜€๋‹ค. ๊ณ„์ˆ˜์ถ”์ •์˜ ๋˜ ๋‹ค๋ฅธ ๋ฐฉ๋ฒ•์œผ๋กœ ์ด์šฉ๋  ์ˆ˜ ์žˆ๋Š” Random Draws๋„ ์žˆ์ง€๋งŒ ๊ณ„์ˆ˜๊ฐ’์„ ์ˆ˜๋ ดํ•˜๊ธฐ ์œ„ํ•ด Halton Draws ๋ฐฉ๋ฒ•์— ๋น„ํ•ด ๋งŽ์€ Draws๊ฐ€ ํ•„์š”ํ•˜์—ฌ ํšจ์œจ์ ์ด์ง€ ๋ชปํ•˜๋‹ค๊ณ  ์•Œ๋ ค์ ธ ์žˆ๋‹ค(Train, 2003).

๊ทธ๋ฆฌ๊ณ  ์ด ์—ฐ๊ตฌ์—์„œ ์ œ์‹œํ•˜๋Š” ํ™•๋ฅ ์  ๋ชจ์ˆ˜๋ฅผ ๊ณ ๋ คํ•œ ์Œ์ดํ•ญ ๋ชจํ˜•์˜ ๊ณ„์ˆ˜์ถ”์ • ์‹œ, ์—ฌ๋Ÿฌ ํ˜•ํƒœ์˜ ํ™•๋ฅ ๋ถ„ํฌ(์ •๊ทœ, ๋กœ๊ทธ์ •๊ทœ, ๊ท ์ผ๋ถ„ํฌ)๋ฅผ ๊ณ ๋ คํ•  ์ˆ˜ ์žˆ๋Š”๋ฐ, ์ด์ค‘ ์ •๊ทœ๋ถ„ํฌ๊ฐ€ ํ†ต๊ณ„์ ์œผ๋กœ ๊ฐ€์žฅ ์œ ํšจํ•œ ๊ฐ’์„ ๋ณด์ด๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค.

๋„์ถœ๋œ ๊ณ„์ˆ˜์˜ ํ•œ๊ณ„ํšจ๊ณผ(Marginal Effect)๊ฐ€ Table 4์— ์ œ์‹œ๋˜์–ด์žˆ๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ๊ธฐํ•˜๊ตฌ์กฐ์˜ ๋‹จ์œ„(unit)๋ณ€ํ™”๊ฐ€ ์‚ฌ๊ณ ๋ฐœ์ƒ์— ๋ฏธ์น˜๋Š” ์ƒ๋Œ€์ ์ธ ์˜ํ–ฅ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ๊ทธ๋ฆฌ๊ณ  ์ด์šฉ๋œ ๋ณ€์ˆ˜์˜ ํŠน์„ฑ์— ๋”ฐ๋ผ์„œ ํƒ„๋ ฅ์„ฑ(Elsaticity)์„ ์ œ์‹œํ•˜๊ณ ์ž ํ•œ๋‹ค.

๋„์ถœ๋œ ๋ชจํ˜•์˜ ์ „์ฒด์ ์ธ ์„ค๋ช…๋ ฅ์€ ๋กœ๊ทธ-์šฐ๋„ํ•จ์ˆ˜ ๊ฐ’์ด -105,377.5์—์„œ -28,315.69๋กœ ์šฐ๋„๋น„(ฯ2)๊ฐ€ 0.731๋กœ ์ผ๋ฐ˜์ ์ธ ์Œ์ดํ•ญ๋ชจํ˜•(-29,210.79, ฯ2=0.722)์— ๋น„ํ•ด ์ƒ๋Œ€์ ์œผ๋กœ ์กฐ๊ธˆ ๋‚˜์€ ์„ค๋ช…๋ ฅ์„ ๊ฐ€์ง€๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค.

์•ž์„œ ์„ค๋ช…ํ•œ๋ฐ๋กœ ํ™•๋ฅ ์  ๋ชจ์ˆ˜๋ฅผ ์ด์šฉํ•œ ์Œ์ดํ•ญ ๋ชจํ˜•์˜ ๊ฒฝ์šฐ, ์ถ”์ •๋œ ๊ณ„์ˆ˜์˜ ํ‘œ์ค€ํŽธ์ฐจ๊ฐ€ ํ†ต๊ณ„์ ์œผ๋กœ ์œ ์˜ํ•œ ๊ฒฝ์šฐ(s.dโ‰ 0, t-value โ‰ฅ1.96)์— ์‚ฌ์šฉ๋  ์ˆ˜ ์žˆ์œผ๋ฉฐ(RPM), ๋งŒ์•ฝ ์ถ”์ •๋œ ๊ณ„์ˆ˜์˜ ํ‘œ์ค€ํŽธ์ฐจ๊ฐ€ ํ†ต๊ณ„์ ์œผ๋กœ ์œ ์˜ํ•˜์ง€ ์•Š์œผ๋ฉด(s.d=0, t-value<1.96) ์ถ”์ •๋œ ๊ณ„์ˆ˜๊ฐ€ ์ผ๋ฐ˜์ ์ธ ์Œ์ดํ•ญ ๋ชจํ˜•์˜ ๊ฒฝ์šฐ์ฒ˜๋Ÿผ, ๊ตฌ๊ฐ„์— ๊ด€๊ณ„์—†์ด ๋ชจ๋“  ๊ตฌ๊ฐ„์— ๋Œ€ํ•ด์„œ ๋™์ผํ•œ ๊ฐ’์„ ๊ฐ€์ง€๊ฒŒ ๋œ๋‹ค(FPM).

Table 3์— ๋”ฐ๋ฅด๋ฉด, ๊ตฌ๊ฐ„์˜ ๊ธธ์ด๊ฐ€ ๊ธธ์–ด์ง€๊ฒŒ ๋˜๋ฉด ์‚ฌ๊ณ ๋ฐœ์ƒํ™•๋ฅ ์ด ๋†’์•„์ง์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. RPM์—์„œ๋Š” ํ‰๊ท ๊ฐ’์ด 0.787 ํ‘œ์ค€ํŽธ์ฐจ๋Š” 0.087์ด๋ฉฐ t-๊ฐ’์— ์˜ํ•˜๋ฉด ์ƒ๋‹นํžˆ ๋†’์€ ์œ ์˜ํ•จ์„ ๋ณด์ž„์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. RPM์˜ ํƒ„๋ ฅ์„ฑ ์ธก๋ฉด์—์„œ 1%์˜ ๊ตฌ๊ฐ„๊ธธ์ด ์ฆ๊ฐ€๋Š” 0.787%์˜ ์‚ฌ๊ณ ์ฆ๊ฐ€์— ์˜ํ–ฅ์„ ์ฃผ๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค. ๊ทธ๋ฆฌ๊ณ  FPM์—์„œ๋Š” 0.743%์˜ ์˜ํ–ฅ์ด ์žˆ๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค. ์ด๋Š” ๋…ธ์ถœ๋Ÿ‰์ด ๋งŽ์•„์งˆ์ˆ˜๋ก ์‚ฌ๊ณ ๋ฐœ์ƒํ™•๋ฅ ์ด ๋†’์•„์ง„๋‹ค๋Š” ๊ธฐ์กด์˜ ์—ฐ๊ตฌ์™€ ์ผ์น˜ํ•จ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ๋˜ํ•œ, ๊ตฌ๊ฐ„ ๊ธธ์ด์˜ ๊ฒฝ์šฐ ํ†ต๊ณ„์ ์œผ๋กœ ์œ ์˜ํ•œ ํ‘œ์ค€์˜ค์ฐจ ๊ฐ’์„ ๋ณด์ด๊ธฐ์— ์‚ฌ๊ณ ์— ๋Œ€ํ•œ ์˜ํ–ฅ์€ ๊ตฌ๊ฐ„์— ๋”ฐ๋ผ ๋‹ค๋ฅด๊ฒŒ ๋‚˜ํƒ€๋‚  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์ตœ๋Œ€ 1168๊ฐœ์˜ ๋‹ค๋ฅธ ๊ฐ’์„ ๊ฐ€์ง€๊ฒŒ ๋œ๋‹ค.

๊ตํ†ต๋Ÿ‰ ๋˜ํ•œ ๊ตฌ๊ฐ„์˜ ๊ธธ์ด์™€ ๋งˆ์ฐฌ๊ฐ€์ง€๋กœ ๊ตํ†ต๋Ÿ‰์ด ์ฆ๊ฐ€ํ• ์ˆ˜๋ก ์‚ฌ๊ณ ๋ฐœ์ƒ์— ์˜ํ–ฅ์„ ๋งŽ์ด ์ฃผ๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค. FPM์—์„œ๋Š” 1%์˜ ๊ตํ†ต๋Ÿ‰ ์ฆ๊ฐ€๊ฐ€ 1.143%์˜ ์‚ฌ๊ณ  ์ฆ๊ฐ€๋กœ ์ด์–ด์ง์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ๋ฐ˜๋ฉด, RPM์—์„œ๋„ ์ƒ๋‹นํžˆ ์œ ์˜ํ•œ 1.036์˜ ํ‰๊ท ๊ฐ’๊ณผ 0.008์˜ ํ‘œ์ค€ํŽธ์ฐจ๊ฐ’์„ ๊ฐ€์ง€๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค. ํƒ„๋ ฅ์„ฑ ์ธก๋ฉด์—์„œ ๋‘ ๊ฐ€์ง€ ๊ฒฝ์šฐ์—์„œ ๋ชจ๋‘ ํƒ„๋ ฅ์ ์ž„์„ ๋ณด์ด๋Š”๋ฐ, 1%์˜ ๊ตํ†ต๋Ÿ‰ ์ฆ๊ฐ€๋Š” 1.143%(FPM)์™€ 1.036%(RPM)์˜ ์‚ฌ๊ณ ์ฆ๊ฐ€์— ์˜ํ–ฅ์„ ๋ฏธ์น˜๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค. ๊ตฌ๊ฐ„์˜ ๊ธธ์ด์˜ ๊ฒฝ์šฐ์™€ ๋งˆ์ฐฌ๊ฐ€์ง€๋กœ ๋…ธ์ถœ๋Ÿ‰์ด ์ฆ๊ฐ€ํ• ์ˆ˜๋ก ์‚ฌ๊ณ ์ฆ๊ฐ€์—๋„ ์˜ํ–ฅ์„ ๋ฏธ์น˜๋Š” ๊ฒƒ์ž„์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ํ‘œ์ค€ํŽธ์ฐจ์—์„œ ์œ ์˜ํ•œ ๊ฐ’์„ ๋‚˜ํƒ€๋‚ด๊ธฐ ๋•Œ๋ฌธ์— ์ตœ๋Œ€ 1168๊ตฌ๊ฐ„์—์„œ ๊ตํ†ต์‚ฌ๊ณ ์— ๊ฐ๊ธฐ ๋‹ค๋ฅธ ์˜ํ–ฅ(%)์„ ๋ฏธ์น˜๋Š” ๊ฒƒ์œผ๋กœ ๋ณผ ์ˆ˜ ์žˆ๋‹ค.

์ธํ„ฐ์ฒด์ธ์ง€ ๊ตฌ๊ฐ„์˜ ๊ฒฝ์šฐ, ๋ณธ์„ ๊ตฌ๊ฐ„์— ๋น„ํ•ด ์‚ฌ๊ณ ๋ฐœ์ƒ์ด ์ž‘๊ฒŒ ๋ฐœ์ƒํ•จ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค.

์ฐจ์„ ์ˆ˜์˜ ๊ฒฝ์šฐ์—๋Š” 2์ฐจ์„ ์„ ๊ฐ€์ง„ ๊ตฌ๊ฐ„์„ ์ œ์™ธํ•œ ๋‚˜๋จธ์ง€ ๊ตฌ๊ฐ„์—์„œ ํ†ต๊ณ„์ ์œผ๋กœ ์œ ์˜ํ•œ ๊ฒฐ๊ณผ๋ฅผ ๋ณด์˜€๋‹ค. ์šฐ์„  ๊ตฌ๊ฐ„๋‚ด์—์„œ ์ฐจ์„ ์ˆ˜๊ฐ€ ๋งŽ์•„์งˆ์ˆ˜๋ก ์‚ฌ๊ณ ๋ฐœ์ƒ๋„ ์ฆ๊ฐ€ํ•จ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ์ด๋Š” ๊ตฌ๊ฐ„์˜ ๊ธธ์ด ๋ฐ ๊ตํ†ต๋Ÿ‰๊ณผ ํ•จ๊ป˜ ๋…ธ์ถœ๋Ÿ‰์ด ์ฆ๊ฐ€ํ•˜๋Š” ์ธก๋ฉด์—์„œ ์‚ฌ๊ณ ๋ฐœ์ƒ์—๋„ ์˜ํ–ฅ์„ ๋ฏธ์น˜๋Š” ๊ฒƒ์œผ๋กœ ํŒ๋‹จ๋œ๋‹ค. ๊ทธ๋ฆฌ๊ณ , ํ•œ๊ณ„ํšจ๊ณผ์˜ ๊ด€์ ์—์„œ๋Š” 4์ฐจ์„ ์—์„œ 5์ฐจ์„  ๋น„์œจ์— ๋”ฐ๋ผ ๊ทธ ์ฆ๊ฐ€ํšจ๊ณผ๊ฐ€ ์ƒ๋Œ€์ ์œผ๋กœ 3์ฐจ์„ ์—์„œ 4์ฐจ์„ ์œผ๋กœ์˜ ์ฆ๊ฐ€ ๋น„์œจ์— ๋น„ํ•ด ์ž‘๊ฑฐ๋‚˜ ๋ฏธ๋น„ํ•จ์„ ๋ณด์—ฌ ๊ตฌ๊ฐ„๋‚ด 4์ฐจ์„ ์˜ ๋น„์œจ์ด ๊ตํ†ต์•ˆ์ „ ๋ฐ ํˆฌ์ž๋น„์šฉ์˜ ์ธก๋ฉด์—์„œ ํšจ์œจ์ ์ธ ๊ฒƒ์œผ๋กœ ํŒ๋‹จ๋œ๋‹ค.

์™ผ์ชฝ ๊ธธ์–ด๊นจ ํญ์˜ ๊ฒฝ์šฐ, 2-ft(60cm)๋ฏธ๋งŒ์˜ ๊ธธ์–ด๊นจ ํญ์„ ์ œ์™ธํ•œ ๋‚˜๋จธ์ง€์—์„œ ์œ ์˜ํ•œ ๊ฒฐ๊ณผ๊ฐ€ ๋„์ถœ๋˜์—ˆ์œผ๋ฉฐ, ๊ธธ์–ด๊นจ ํญ์ด ๋„“์–ด์งˆ์ˆ˜๋ก ๊ตํ†ต์‚ฌ๊ณ ๊ฐ€ ๊ฐ์†Œํ•จ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค.

3-4ft(90~120cm)์™€ 10ft(300cm)์ด์ƒ์˜ ๊ฒฝ์šฐ์—์„œ๋Š” RPM์˜ ํ‘œ์ค€ํŽธ์ฐจ๊ฐ€ ์œ ์˜ํ•จ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ์šฐ์„  3-4ft๋Š” 78.9%์˜ ๊ตฌ๊ฐ„์—์„œ ์‚ฌ๊ณ ๊ฑด์ˆ˜์˜ ๊ฐ์†Œ๊ฐ€, ๋‚˜๋จธ์ง€ 21.1%์˜ ๊ตฌ๊ฐ„์—์„  ์‚ฌ๊ณ ๊ฑด์ˆ˜๊ฐ€ ์ฆ๊ฐ€ํ•œ๋‹ค. 10ft ์ด์ƒ์˜ ๊ธธ์–ด๊นจํญ์„ ๊ฐ€์ง„ ๊ฒฝ์šฐ์—๋Š” 97%์˜ ๊ตฌ๊ฐ„์—์„œ ์‚ฌ๊ณ ๊ฑด์ˆ˜์˜ ๊ฐ์†Œ๊ฐ€, 3%์˜ ๊ตฌ๊ฐ„์—์„œ ์‚ฌ๊ณ ๊ฑด์ˆ˜๊ฐ€ ์ฆ๊ฐ€ํ•˜๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚˜, ์šด์ „์ž์˜ ๋ถ€์ฃผ์˜ ํ˜น์€ ์ฐจ๋Ÿ‰์˜ ๋ฌธ์ œ๋“ฑ์œผ๋กœ ์ธํ•œ ์ฐจ์„ ์ดํƒˆ๋“ฑ์˜ ์˜ˆ์ƒ์น˜ ๋ชปํ•œ ์ƒํ™ฉ์˜ ๊ฒฝ์šฐ์—๋„ ์ด๋ฅผ ๋ฐฉ์ง€ํ• ์ˆ˜ ์žˆ๋Š” ์ถฉ๋ถ„ํ•œ ํญ์˜ ์™„์ถฉ์ง€์—ญ๊ฐ€ ์‚ฌ๊ณ ๋ฐœ์ƒ๊ฑด์ˆ˜ ๊ฐ์†Œ์— ๋งŽ์€ ๊ธฐ์—ฌํ•จ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ํ•˜์ง€๋งŒ, ํ•œ๊ณ„ํšจ๊ณผ์ธก๋ฉด์—์„  10ft ์ด์ƒ์˜ ๊ฒฝ์šฐ๊ฐ€ 5-9ft(150~270cm)๊ฒฝ์šฐ๋ณด๋‹ค ์ƒ๋Œ€ํšจ๊ณผ๊ฐ€ ์ค„์–ด๋“œ๋Š” ๊ฒƒ์œผ๋กœ ๋ณด์ด๋Š”๋ฐ, ์‚ฌํšŒ์  ๋น„์šฉ๋“ฑ์˜ ๊ฒฝ์ œ์  ์ธก๋ฉด์—์„œ 5-9ft์˜ ๊ธธ์–ด๊นจ ํญ์„ ๊ฐ€์ง„ ๊ธธ์–ด๊นจ๊ฐ€ ๋”์šฑ ํšจ์œจ์ ์ž„์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค.

์˜ค๋ฅธ์ชฝ ๊ธธ์–ด๊นจ ํญ์˜ ๊ฒฝ์šฐ๋„, ์™ผ์ชฝ ๊ธธ์–ด๊นจ ํญ๊ณผ ๋งˆ์ฐฌ๊ฐ€์ง€๋กœ 2-ft ๊ตฌ๊ฐ„์€ ์œ ์˜ํ•˜์ง€ ์•Š์€ ๊ฒฐ๊ณผ๋ฅผ ๋ณด์ด๋Š” ๋ฐ˜๋ฉด, ๋‚˜๋จธ์ง€ ๊ฒฝ์šฐ์—์„œ๋Š” ์œ ์˜ํ•œ ๊ฒฐ๊ณผ๊ฐ€ ๋„์ถœ๋˜์–ด ์˜ค๋ฅธ์ชฝ ๊ธธ์–ด๊นจ์˜ ํญ์ด ๋„“์–ด์งˆ์ˆ˜๋ก ์‚ฌ๊ณ ๋ฐœ์ƒ์€ ์ค„์–ด๋“œ๋Š” ์‚ฌ์‹ค์„ ๋ณด์ธ๋‹ค. 5-9ftํญ์˜ ๊ฒฝ์šฐ์—์„œ ํ‰๊ท ๊ฐ’๊ณผ ํ‘œ์ค€ํŽธ์ฐจ๊ฐ€ ํ†ต๊ณ„์ ์œผ๋กœ ์œ ์˜ํ•จ์„ ๋ณด์—ฌ, ๊ฐ ๊ตฌ๊ฐ„๋งˆ๋‹ค ๋‹ค๋ฅธ ๊ฐ’์„ ๊ฐ€์ง์„ ์•Œ ์ˆ˜ ์žˆ์œผ๋ฉฐ, ๊ฑฐ์˜ ๋Œ€๋ถ€๋ถ„์˜ ๊ตฌ๊ฐ„(99.6%)์—์„œ ์‚ฌ๊ณ ๊ฑด์ˆ˜๊ฐ€ ๊ฐ์†Œํ•˜๋Š” ๊ฒƒ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ํ•œ๊ณ„ํšจ๊ณผ ์ธก๋ฉด์—์„œ๋Š” ์™ผ์ชฝ ๊ธธ์–ด๊นจ ํญ์˜ ๊ฒฝ์šฐ์™€ ๋‹ฌ๋ฆฌ 10-ft์ด์ƒ์˜ ๊ธธ์–ด๊นจ ํญ์„ ๊ฐ€์ง„ ๊ตฌ๊ฐ„์—์„œ๋Š” ์‚ฌ๊ณ ๊ฐ์†Œ์— ๋Œ€ํ•œ ํšจ๊ณผ๊ฐ€ ๋Š˜์–ด๋‚˜๊ฑฐ๋‚˜(FPM์˜ ๊ฒฝ์šฐ), ์ƒ๋Œ€์  ํšจ๊ณผํฌ๊ธฐ๊ฐ€ ์™ผ์ชฝ ๊ธธ์–ด๊นจ์˜ ๊ฒฝ์šฐ๋ณด๋‹ค ์กฐ๊ธˆ ์ค„์–ด๋“ฆ(RPM์˜ ๊ฒฝ์šฐ)์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ๋„๋กœ์˜ ์™ผ์ชฝ ๊ธธ์–ด๊นจ ํญ๊ณผ ๋‹ฌ๋ฆฌ ์˜ค๋ฅธ์ชฝ ๊ธธ์–ด๊นจ ํญ์—๋Š” ์†๋„์ œํ•œ๋“ฑ์˜ ํ‘œ์ง€ํŒ, ์‹œ์„ค๋ฌผ ํ˜น์€ ๋‹จ์†์„ ์œ„ํ•œ ์ˆœ์ฐฐ์ฐจ๋“ฑ์˜ ์ถฉ๋Œ์‚ฌ๊ณ ๋ฅผ ์œ ๋ฐœ์‹œํ‚ฌ ์ˆ˜ ์žˆ๋Š” ์ธ์ž๊ฐ€ ์กด์žฌํ•˜๊ธฐ ๋•Œ๋ฌธ์— ๋„“์€ ์–ด๊นจํญ์ด ์‚ฌ๊ณ ๊ฐ์†Œ ํšจ๊ณผ์— ํฐ ์˜ํ–ฅ์„ ๋ฏธ์น ๊ฒƒ์œผ๋กœ ์‚ฌ๋ฃŒ๋œ๋‹ค.

Table 4. Average Marginal Effects for Fixed-and Random- Parameters Negative Binomial Models

Fixed Parameter Model

Random Parameter Model

LNLEN

7.788

4.251

LNADT

11.971

5.597

INTERCHANGE

-0.287

-0.275

NLN3

3.932

2.556

NLN4

9.454

4.625

NLN5

8.219

5.429

LSHW34

-2.858

-1.245

LSHW59

-5.193

-2.125

LSHW10

-4.359

-1.623

RSHW34

-2.998

-1.504

RSHW59

-3.343

-1.860

RSHW10

-4.134

-1.726

NHORZ

0.302

0.105

MINHLMI

-6.544

-2.777

MAXHDEL

0.066

0.026

MAXVCRVG

0.208

0.156

๊ตฌ๊ฐ„๋‚ด์— ์กด์žฌํ•˜๋Š” ํšก๋‹จ๊ณก์„ ์˜ ๊ฐœ์ˆ˜์˜ ๊ฒฝ์šฐ, ๊ณก์„ ์˜ ๊ฐœ์ˆ˜๊ฐ€ ์ฆ๊ฐ€ํ• ์ˆ˜๋ก ์‚ฌ๊ณ ๊ฑด์ˆ˜๋„ ์ฆ๊ฐ€ํ•จ์„ ๋ณด์—ฌ์ค€๋‹ค. ๋ฐ˜๋ฉด, ์ข…๋‹จ ๊ณก์„ ์˜ ๊ฐœ์ˆ˜๋Š” ํ†ต๊ณ„์ ์œผ๋กœ ์œ ์˜ํ•˜์ง€ ์•Š์Œ์„ ๋‚˜ํƒ€๋‚ด์–ด ๊ตํ†ต์‚ฌ๊ณ ๋ฐœ์ƒ์—๋Š” ํšก๋‹จ๊ณก์„ ์˜ ๊ฒฝ์šฐ๊ฐ€ ๋” ๋งŽ์€ ์˜ํ–ฅ์„ ์ฃผ๋Š” ๊ฒƒ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค.

์งง์€ ํšก๋‹จ๊ณก์„ ์˜ ๊ธธ์ด๊ฐ€ ๊ธธ์–ด์ง์— ๋”ฐ๋ผ ์‚ฌ๊ณ ๋ฐœ์ƒ์„ ๊ฐ์†Œ์‹œํ‚ค๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค. ๋ฌผ๋ก  ๊ตฌ๊ฐ„๋‚ด์—์„œ 1-mile(1.609km) ๊ธธ์ด์˜ ํšก๋‹จ๊ณก์„ ์˜ ๋‹จ์œ„์ฆ๊ฐ€๋Š” ํ˜„์‹ค์ ์œผ๋กœ ๋ถˆ๊ฐ€๋Šฅํ•˜์ง€๋งŒ, ๊ทธ์— ๋”ฐ๋ฅธ ํ•œ๊ณ„ํšจ๊ณผ ์ธก๋ฉด์—์„œ๋Š” ๊ตํ†ต์‚ฌ๊ณ ์— ๋Œ€ํ•œ ๊ฐ์†Œํšจ๊ณผ๋ฅผ ๋ณด์—ฌ์ค€๋‹ค.

๊ตฌ๊ฐ„๋‚ด์— ์กด์žฌํ•˜๋Š” ํšก๋‹จ๊ณก์„ ์˜ ์ตœ๋Œ€๊ต๊ฐ์˜ ๊ฒฝ์šฐ, ๊ต๊ฐ์˜ ํฌ๊ธฐ๊ฐ€ ์ฆ๊ฐ€ํ• ์ˆ˜๋ก ๊ตํ†ต์‚ฌ๊ณ  ๋ฐœ์ƒ์— ์˜ํ–ฅ์„ ์ฃผ๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค. ์ด๋Š” ํšก๋‹จ๊ณก์„ ๊ตฌ๊ฐ„์˜ ๊ต๊ฐ์ด ์ฆ๊ฐ€ํ• ์ˆ˜๋ก ๊ณก์„ ๊ตฌ๊ฐ„์˜ ๋„๋กœ ์„ ํ˜•์˜ ํ˜•ํƒœ๊ฐ€ ์ง์„ ํ™”๋จ์— ๋”ฐ๋ฅธ ์ฐจ๋Ÿ‰ ์†๋„์˜ ์ฆ๊ฐ€์— ๋”ฐ๋ฅธ ์˜ํ–ฅ์œผ๋กœ ํŒ๋‹จ๋œ๋‹ค. ๋”ฐ๋ผ์„œ, ๋„๋กœ๊ตฌ๊ฐ„๋‚ด์—์„œ ํšก๋‹จ๊ณก์„ ์˜ ์ตœ๋Œ€๊ต๊ฐ๋ฐœ์ƒ์„ ์ตœ์†Œํ™”ํ•˜๋Š” ๋ฐฉํ–ฅ์˜ ์„ค๊ณ„๊ฐ€ ํ•„์š”ํ•  ๊ฒƒ์ด๋‹ค.

์ตœ๋Œ€์ข…๋‹จ๊ตฌ๋ฐฐ์˜ ๊ฒฝ์šฐ, ์ ˆ๋Œ€๊ฐ’์œผ๋กœ ์‚ฐ์ถœ๋˜์—ˆ์œผ๋ฉฐ, ๊ตฌ๋ฐฐ๊ฐ€ ์ฆ๊ฐ€ํ•  ์ˆ˜๋ก ์‚ฌ๊ณ ๊ฑด์ˆ˜๋„ ์ฆ๊ฐ€ํ•จ์„ ๋ณด์—ฌ์ค€๋‹ค. ์ข…๋‹จ๊ตฌ๋ฐฐ๊ฐ€ 1หš์ฆ๊ฐ€ํ•จ์— ๋”ฐ๋ผ์„œ ์‚ฌ๊ณ ๋ฐœ์ƒ์—๋Š” FPM์—์„œ๋Š” 0.208, RPM์—์„œ๋Š” 0.156์˜ ์ฆ๊ฐ€์— ์˜ํ–ฅ์„ ๋ฏธ์น˜๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค.

4.3 ๋ชจํ˜•๋น„๊ต

FPM ๋ชจํ˜•๊ณผ RPM ๋ชจํ˜•์˜ ์ถ”์ •๊ฐ’ ๋น„๊ต๋ฅผ ์œ„ํ•ด 3๊ฐ€์ง€ ๋ฐฉ๋ฒ•์œผ๋กœ ์ถ”์ •๊ฐ’์„ ๋น„๊ตํ•˜์˜€๋‹ค. RMSE๋ฅผ ์ œ์™ธํ•œ ์ถ”์ •์—์„œ๋Š” ๋ฏธ๋น„ํ•˜์ง€๋งŒ ๋ณด๋‹ค ๋‚˜์€ ์ถ”์ •๊ฐ’์„ ๋ณด์ž„์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค.

Table 5. Prediction Accuracy

FPM

RPM

RMSE (Root Mean Seuare Mean)

18.01

13.13

MAPE (Mean Absolute Perchange Error)

0.00332

0.00322

MAE (Mean Absolute Error)

5.473

5.429

5. ๊ฒฐ๋ก  ๋ฐ ํ–ฅํ›„ ์—ฐ๊ตฌ๊ณผ์ œ

์ด๋ฒˆ ์—ฐ๊ตฌ์—์„œ๋Š” ๊ธฐ์กด์— ์‚ฌ์šฉ๋˜๋˜ ์ผ๋ฐ˜์ ์ธ ์Œ์ดํ•ญ ๋ชจํ˜•์—์„œ ๋ณด๋‹ค ๋ฐœ์ „๋œ ํ˜•ํƒœ์˜ ํ™•๋ฅ ์  ๋ชจ์ˆ˜๋ฅผ ๊ณ ๋ คํ•œ ์Œ์ดํ•ญ ๋ชจํ˜•์„ ์ œ์‹œํ•˜์—ฌ ๊ตํ†ต์‚ฌ๊ณ ์™€ ๊ธฐํ•˜๊ตฌ์กฐ์™€์˜ ๊ด€๊ณ„๋ฅผ ํŒŒ์•…ํ•˜์˜€๋‹ค. ๊ธฐ์กด์˜ ์Œ์ดํ•ญ ๋ชจํ˜•์€ ๊ฐ ๋„๋กœ๊ตฌ๊ฐ„์— ์„ค์น˜๋œ ๊ธฐํ•˜๊ตฌ์กฐ์˜ ์ข…๋ฅ˜ ๋ฐ ์„ค์น˜ํ˜•ํƒœ๊ฐ€ ๋™์ผํ•˜์ง€ ์•Š์Œ์—๋„ ๋ถˆ๊ตฌํ•˜๊ณ , ๊ตฌ๊ฐ„์˜ ์ด์งˆ์„ฑ์„ ๊ณ ๋ คํ•˜์ง€ ๋ชปํ•œ ๋‹จ์ ์œผ๋กœ ์ธํ•ด ์‚ฌ๊ณ ์˜ˆ์ธก์‹œ ์‚ฌ๊ณ ์˜ ์˜ˆ์ธก๊ฐ’์— ๋Œ€ํ•œ ๋ถˆํ™•์‹ค์„ฑ(uncertainity)์™€ ์ž„์˜์„ฑ(randomness)์„ ๋‚ดํฌํ•˜๊ณ  ์žˆ์—ˆ๋‹ค. ์ด๋Ÿฌํ•œ ์ด์œ ๋กœ, ๊ฐœ์„ ์ด ํ•„์š”ํ•œ ๊ตฌ๊ฐ„์ด ํˆฌ์ž์šฐ์„ ์ˆœ์œ„์—์„œ ๋ฐ€๋ ค๋‚˜, ์ ์ ˆํ•œ ์˜ˆ์‚ฐ์ด ๋ถ„๋ฐฐ๋˜์ง€ ๋ชปํ•˜์—ฌ ๊ฐœ์„  ์‚ฌ์—…์ด ์ œ๋•Œ ์ด๋ฃจ์–ด์ง€์ง€ ๋ชปํ•˜๋Š” ์›์ธ์ด ๋˜๊ธฐ๋„ ํ–ˆ์—ˆ๋‹ค.

์ด๋ฅผ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•œ ๋ฐฉ๋ฒ•์œผ๋กœ ์ด ์—ฐ๊ตฌ์—์„œ๋Š” ๊ตฌ๊ฐ„์˜ ์ด์งˆ์„ฑ์„ ๊ณ ๋ คํ•˜์˜€์œผ๋ฉฐ, ๋„์ถœ๋œ ๋ชจํ˜•์˜ ๊ฒฐ๊ณผ๋Š” ๋น„๊ต๋ฅผ ์œ„ํ•ด ๊ธฐ์กด์— ์‚ฌ์šฉ๋˜์—ˆ๋˜ ์ผ๋ฐ˜์ ์ธ ์Œ์ดํ•ญ๊ณผ ํ•จ๊ป˜ ์ œ์‹œ๋˜์—ˆ๋‹ค. ๋ชจํ˜•์˜ ์„ค๋ช…๋ ฅ์€ ๊ธฐ์กด์˜ ์Œ์ดํ•ญ๋ชจํ˜•์— ๋น„ํ•ด ๋†’์€ ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค. ์ด 16๊ฐœ์˜ ๊ตํ†ต๋Ÿ‰, ๊ตฌ๊ฐ„ ๊ธธ์ด ๋ฐ ๋‹ค์–‘ํ•œ ๊ธฐํ•˜๊ตฌ์กฐ(์ฐจ์„ ์ˆ˜, ์ขŒโ€ค์šฐ ๊ธธ์–ด๊นจ ํญ, ์ข…โ€คํšก๋‹จ ๊ณก์„ )์™€ ๊ด€๋ จ๋œ ๋ณ€์ˆ˜๋“ค์ด ์‚ฌ์šฉ๋˜์—ˆ๋‹ค. ์ด ์ค‘, 8๊ฐœ์˜ ๋ณ€์ˆ˜๊ฐ€ ๊ฐ ๊ตฌ๊ฐ„๋งˆ๋‹ค ๋‹ค๋ฅธ ํŠน์„ฑ์„ ๋‚˜ํƒ€๋‚ด๋Š” ์ด์งˆ์„ฑ์„ ๊ฐ€์ง€๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ์œผ๋ฉฐ, ๋‚˜๋จธ์ง€ 8๊ฐœ์˜ ๋ณ€์ˆ˜๋Š” ๊ฐ ๊ตฌ๊ฐ„์— ๊ด€๊ณ„์—†์ด ๋™์ผํ•œ ํŠน์„ฑ์„ ๋‚˜ํƒ€๋‚˜๋Š” ๋ณ€์ˆ˜๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค. ๊ตฌ๊ฐ„์˜ ๊ธธ์ด, ๊ตํ†ต๋Ÿ‰, ์ฐจ์„ ์ˆ˜(3, 4์ฐจ์„ ), ์™ผ์ชฝ ๊ธธ์–ด๊นจํญ(3-4, 10ft์ด์ƒ - 90~120, 300cm์ด์ƒ), ์˜ค๋ฅธ์ชฝ ๊ธธ์–ด๊นจํญ(5-9ft - 150~270cm), ๊ตฌ๊ฐ„๋‚ด ์ข…๋‹จ๊ณก์„ ์˜ ์ตœ๋Œ€๊ตฌ๋ฐฐ๊ฐ€ ์ด์งˆ์„ฑ์„ ๊ฐ€์ง€๋Š” ๋ณ€์ˆ˜๋กœ ๋„์ถœ๋˜์—ˆ๋‹ค. ์ฐจ์„ ์ˆ˜์˜ ๊ฒฝ์šฐ, 4์ฐจ์„ ์—์„œ 5์ฐจ์„ ์ด์ƒ์œผ๋กœ ๋ณ€ํ•˜๋Š” ๊ฒฝ์šฐ, ์‚ฌ๊ณ ๊ฐ์†Œ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์ด ์ƒ๋Œ€์ ์œผ๋กœ ์ค„์–ด๋“œ๋Š” ๊ฒƒ์„ ์•Œ ์ˆ˜ ์žˆ์—ˆ๋Š”๋ฐ, ์ด๋Š” ์ฐจ์„ ์ˆ˜์˜ ์ฆ๊ฐ€์— ๋”ฐ๋ฅธ ๋น„์šฉ๊ณผ ํ•จ๊ป˜ ๊ทธ์— ๋”ฐ๋ฅธ ์‚ฌ๊ณ ๋น„์šฉ์ธก๋ฉด์—์„œ ๋ณด๋‹ค ์ž์„ธํ•œ ์—ฐ๊ตฌ๊ฐ€ ํ•„์š”ํ•  ๊ฒƒ์œผ๋กœ ์‚ฌ๋ฃŒ๋œ๋‹ค. ์ด๋Ÿฌํ•œ ํŠน์„ฑ์€ ๊ธธ์–ด๊นจ ํญ์—์„œ๋„ ๋‚˜ํƒ€๋‚ฌ๋Š”๋ฐ, ์ฐจ์„ ์ˆ˜์™€ ๋งˆ์ฐฌ๊ฐ€์ง€๋กœ ๋น„์šฉ-์ด์ต์ธก๋ฉด์—์„œ์˜ ์—ฐ๊ตฌ๊ฐ€ ํ•„์š”ํ•  ๊ฒƒ์ด๋‹ค. ๋˜ํ•œ, ๋„๋กœ๊ฑด์„ค ํ˜น์€ ๊ธฐํ•˜๊ตฌ์กฐ์˜ ๋ณ€๊ฒฝ ์‹œ, ์ž์—ฐํ™˜๊ฒฝ์˜ ์ œ์•ฝ ๋“ฑ์œผ๋กœ ์ธํ•ด ๊ธฐ์ค€์— ๋ฏธ์น˜์ง€ ๋ชปํ•˜๋Š” ๊ธฐํ•˜๊ตฌ์กฐ๊ฐ€ ๊ฑด์„ค๋˜์–ด์•ผ ํ•  ๊ฒฝ์šฐ, ์ตœ๋Œ€ํ•œ์˜ ์•ˆ์ „์„ฑ์„ ๊ณ ๋ คํ•  ์ˆ˜ ์žˆ๋Š” ๊ธฐํ•˜๊ตฌ์กฐ์˜ ์ง€์นจ์ด ๋  ์ˆ˜ ์žˆ์„ ๊ฒƒ์ด๋‹ค.

์ด ์—ฐ๊ตฌ๋Š” ์•ˆ์ „๋ถ„์•ผ์—์„œ ์œ ์˜ํ•œ ๊ฒฐ๊ณผ๊ฐ€ ๋„์ถœ๋œ๋‹ค๊ณ  ํŒ๋‹จ๋˜๋Š” 5๋…„ ๋™์•ˆ์˜ ์‚ฌ๊ณ ์ž๋ฃŒ๋ณด๋‹ค ํ›จ์”ฌ ์˜ค๋žœ ๊ธฐ๊ฐ„ ๋™์•ˆ์˜ ์‚ฌ๊ณ ์ž๋ฃŒ๊ฐ€ ์ด์šฉ๋˜์–ด ๋ชจํ˜•์˜ ์„ค๋ช…๋ ฅ์ด ๋†’์•„์กŒ์Œ์—๋„ ๋ถˆ๊ตฌํ•˜๊ณ  ๋…๋ฆฝ๋ณ€์ˆ˜๋กœ ์‚ฌ์šฉ๋œ ๊ธฐํ•˜๊ตฌ์กฐ์˜ ๊ฒฝ์šฐ, ๊ทธ ๋ณ€ํ™”๋ฅผ ๋งค๋…„ ์ถ”์ ํ•˜๊ธฐ์—๋Š” ์ž๋ฃŒ์˜ ์ˆ˜์ง‘ ๋“ฑ์— ์–ด๋ ค์›€์ด ์žˆ์—ˆ๋‹ค. ๋˜ํ•œ, ๋‚ ์”จโ€ค์‹œ๊ฐ„ ๋“ฑ์— ๊ด€๋ จ๋œ ๋‹ค๋ฅธ ํ™˜๊ฒฝ์ ์ธ ์š”์†Œ ๋ฐ ์šด์ „์žํŠน์„ฑ(์„ฑ๋ณ„, ๋‚˜์ด, ์Œ์ฃผ ๋“ฑ)์— ๊ด€๋ จ๋œ ์ž๋ฃŒ๊ฐ€ ์ ์šฉ๋˜์ง€ ๋ชปํ•˜์˜€๋‹ค. ์ด ๋ชจ๋“  ์ž๋ฃŒ๊ฐ€ ์ˆ˜์ง‘๋˜๊ธฐ์—๋Š” ๋ณด๋‹ค ๋งŽ์€ ๋น„์šฉ๊ณผ ์‹œ๊ฐ„์ด ์†Œ์š”๋˜๊ฒ ์ง€๋งŒ, ๊ณ„์†์ ์ธ ๋…ธ๋ ฅ์ด ์š”๊ตฌ๋˜๋ฉฐ, ์ด๋Ÿฌํ•œ ์ž๋ฃŒ๊ฐ€ ์ด์šฉ๊ฐ€๋Šฅํ•˜๋‹ค๋ฉด, ๊ตํ†ต์‚ฌ๊ณ ๋ฟ๋งŒ ์•„๋‹ˆ๋ผ, ๋น„์šฉ-ํšจ๊ณผ์  ์ธก๋ฉด์—์„œ์˜ ๊ธฐํ•˜๊ตฌ์กฐ ์ง€์นจ์— ๋Œ€ํ•œ ๋ณด๋‹ค ์ž์„ธํ•˜๊ณ  ์„ธ๋ฐ€ํ•œ ์—ฐ๊ตฌ์— ๋งŽ์€ ๋„์›€์ด ๋  ๊ฒƒ์ด๋‹ค.

Acknowledgements

References

1 
Alfonso, M., Lucio, C. and Renato, L. (2008). โ€œCrash prediction models for rural motorways.โ€ Transportation Research Record: Journal of the Transportation Research Board, No. 2083, pp. 180-189, Transportation Research Board, Washington D.C.Alfonso, M., Lucio, C. and Renato, L. (2008). โ€œCrash prediction models for rural motorways.โ€ Transportation Research Record: Journal of the Transportation Research Board, No. 2083, pp. 180-189, Transportation Research Board, Washington D.C.Google Search
2 
Bhat, C. (2003). โ€œSimulation estimation of mixed discrete choice models using randomized and scrambled Halton sequences.โ€ Transportation Research Part B, Vol. 37, No. 1, pp. 837-855.10.1016/S0191-2615(02)00090-5Bhat, C. (2003). โ€œSimulation estimation of mixed discrete choice models using randomized and scrambled Halton sequences.โ€ Transportation Research Part B, Vol. 37, No. 1, pp. 837-855.DOI
3 
Chin, H. C. and Quddus, M. A. (2003). โ€œApplying the random effect negative binomial model to examine traffic accident occurrence at signalized intersections.โ€ Accident Analysis and Prevention, Vol. 35, pp. 253-259.10.1016/S0001-4575(02)00003-9Chin, H. C. and Quddus, M. A. (2003). โ€œApplying the random effect negative binomial model to examine traffic accident occurrence at signalized intersections.โ€ Accident Analysis and Prevention, Vol. 35, pp. 253-259.DOI
4 
Ciro, C., Maurizio, G. and Parisi, A. (2007). โ€œA crash-prediction model for multilane roads.โ€ Accident Analysis and Prevention, Vol. 39, pp. 657-670.10.1016/j.aap.2006.10.012Ciro, C., Maurizio, G. and Parisi, A. (2007). โ€œA crash-prediction model for multilane roads.โ€ Accident Analysis and Prevention, Vol. 39, pp. 657-670.DOI
5 
Engel, J. (1984). โ€œModels for response data showing extra-poisson variation.โ€ Statistical Neerlandica, Vol. 38, No. 3. pp. 159-167.10.1111/j.1467-9574.1984.tb01107.xEngel, J. (1984). โ€œModels for response data showing extra-poisson variation.โ€ Statistical Neerlandica, Vol. 38, No. 3. pp. 159-167.DOI
6 
Jovanis, P. P. and Chang, H. L. (1986). โ€œModeling the Relationship of Accidents to Miles Traveled.โ€ Transportation Research Record: Journal of the Transportation Research Board, Vol. 1068, pp. 42-51, Transportation Research Board, Washington D.C.Jovanis, P. P. and Chang, H. L. (1986). โ€œModeling the Relationship of Accidents to Miles Traveled.โ€ Transportation Research Record: Journal of the Transportation Research Board, Vol. 1068, pp. 42-51, Transportation Research Board, Washington D.C.Google Search
7 
Lawless, J. (1987). โ€œNegative binomial and mixed poisson regression.โ€ Canadian Journal of Statistics, Vol. 15, No. 3, pp. 209-225.10.2307/3314912Lawless, J. (1987). โ€œNegative binomial and mixed poisson regression.โ€ Canadian Journal of Statistics, Vol. 15, No. 3, pp. 209-225.DOI
8 
Maher, M. (1991). โ€œA new bivariate negative binomial model for accident frequencies.โ€ Traffic Engineering and Control, Vol. 32, No. 9, pp. 422-433.Maher, M. (1991). โ€œA new bivariate negative binomial model for accident frequencies.โ€ Traffic Engineering and Control, Vol. 32, No. 9, pp. 422-433.Google Search
9 
Miaou, S. P. and Lum, H. (1993). โ€œModeling vehicle accidents and highway at geometric design relationships.โ€ Accident Analysis and Prevention Vol. 25, No. 6, pp. 689-709.10.1016/0001-4575(93)90034-TMiaou, S. P. and Lum, H. (1993). โ€œModeling vehicle accidents and highway at geometric design relationships.โ€ Accident Analysis and Prevention Vol. 25, No. 6, pp. 689-709.DOI
10 
Milton, J., Shankar, V. and Mannering, F. (2008). โ€œHighway accident severities and the mixed logit model: an exploratory empirical analysis.โ€ Accident Analysis and Prevention, Vol. 40, No. 1, pp. 260-266.10.1016/j.aap.2007.06.006Milton, J., Shankar, V. and Mannering, F. (2008). โ€œHighway accident severities and the mixed logit model: an exploratory empirical analysis.โ€ Accident Analysis and Prevention, Vol. 40, No. 1, pp. 260-266.DOI
11 
Noland, R. B. (2003). โ€œTraffic fatalities and injuries: The effect of changes in infrastructure and other trends.โ€ Accident Analysis and Prevention, Vol. 35, pp. 599-611.10.1016/S0001-4575(02)00040-4Noland, R. B. (2003). โ€œTraffic fatalities and injuries: The effect of changes in infrastructure and other trends.โ€ Accident Analysis and Prevention, Vol. 35, pp. 599-611.DOI
12 
Ogden, K. W. (1997). โ€œThe effects of paved shoulders on accidents on rural highways.โ€ Accident Analysis and Prevention, Vol. 29, No. 3, pp. 353-362.10.1016/S0001-4575(97)00001-8Ogden, K. W. (1997). โ€œThe effects of paved shoulders on accidents on rural highways.โ€ Accident Analysis and Prevention, Vol. 29, No. 3, pp. 353-362.DOI
13 
Sarath, C. J. and Nicholas, J. G. (1990). โ€œEstimating truck accident rate and involvements using linear and poisson regression models.โ€ Transportation Planning and Technology, Vol. 15, No. 1, pp. 41-58.10.1080/03081069008717439Sarath, C. J. and Nicholas, J. G. (1990). โ€œEstimating truck accident rate and involvements using linear and poisson regression models.โ€ Transportation Planning and Technology, Vol. 15, No. 1, pp. 41-58.DOI
14 
Shankar, V. N., Albin, R. B., Milton, J. C. and Mannering, F. L. (1998). โ€œEvaluating median cross-over likelihoods with clustered accident counts: An empirical inquiry using random effects negative binomial model.โ€ Transportation Research Record: Journal of the Transportation Research Board, No. 1635, pp. 44-48, Transportation Research Board, Washington D.C.Shankar, V. N., Albin, R. B., Milton, J. C. and Mannering, F. L. (1998). โ€œEvaluating median cross-over likelihoods with clustered accident counts: An empirical inquiry using random effects negative binomial model.โ€ Transportation Research Record: Journal of the Transportation Research Board, No. 1635, pp. 44-48, Transportation Research Board, Washington D.C.Google Search
15 
Train, K. (2003). Discrete choice models with simulation, Cambridge University Press, Cambridge, UK.10.1017/CBO9780511753930Train, K. (2003). Discrete choice models with simulation, Cambridge University Press, Cambridge, UK.DOI
16 
Washington, S., Karlaftis, M. and Mannering, F. (2010). Statistical and econometric models for transportation data analysis, Chapman Hall/CRC, Boca Raton, Fla.Washington, S., Karlaftis, M. and Mannering, F. (2010). Statistical and econometric models for transportation data analysis, Chapman Hall/CRC, Boca Raton, Fla.Google Search
17 
Zhang, C. and Ivan, J. N. (2005). โ€œEffects of geometric characteristics on head-on crash incidence on two-lane roads in connecticut.โ€ Transportation Research Record : Journal of the Transportation Research Board, Vol. 1908, pp. 159-164, Transportation Research Board, Washington D.C.Zhang, C. and Ivan, J. N. (2005). โ€œEffects of geometric characteristics on head-on crash incidence on two-lane roads in connecticut.โ€ Transportation Research Record : Journal of the Transportation Research Board, Vol. 1908, pp. 159-164, Transportation Research Board, Washington D.C.Google Search