test driving a small-area population forecasting model: seeking additional horsepower through...
TRANSCRIPT
![Page 1: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/1.jpg)
Test Driving a Small-Area Population Forecasting Model:
Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives
Paul R. Voss and Guangqing ChiApplied Population Laboratory
Center for Demography and EcologyUniversity of Wisconsin – Madison
BSPS Annual Conference 2006
September 2006
The University of Southampton
Support provided by the Wisconsin Agricultural Experiment Station (Hatch project no. WIS04536)
![Page 2: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/2.jpg)
Motivating Questions• What can be done to improve the abysmally atheoretical
nature of small-area population forecasts?• In particular, what about a regression approach?• Especially, what if we step outside our disciplinary confines
and incorporate variables from other fields that, at face value, must be predictors of population growth?
• nature of the land (ground cover, wetlands, hydrography, slope)• accessibility (transportation infrastructure, highways, airports, etc.)• developability (high/low growth potential)• desirability (natural and built amenities)• livibility (potential quality of living)
• And, surely, should we not begin immediately to adopt some of the spatial econometric approaches long effectively employed by quantitative geographers and regional scientists?
![Page 3: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/3.jpg)
• Broaden our thinking regarding the relationships between population change and the host of factors influencing such change – some drawn from demography but many others from disciplines not normally involved in formal population forecasting efforts
• Categorize and integrate these factors in an effective way (construct indexes)
• Incorporate spatial process effects into the model• Carry out the forecasting at a sufficiently fine geogra
phic level that environmental and geophysical effects on population change can be better captured and modeled
Proposed Regression Approach
![Page 4: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/4.jpg)
Strategy
• Assemble all necessary data for 1990 base year
• Forecast populations for 2000
• Compare 2000 forecasts with 2000 census results
![Page 5: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/5.jpg)
Preview of Findings…
It didn’t work
![Page 6: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/6.jpg)
Our Region1,837 minor civil divisions in state of Wisconsin, U.S.
Our Datacensus data
satellite imagery
other data from several federal and state statistical agencies
![Page 7: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/7.jpg)
Population
Demographics
AccessibilityDevelopability
Livability Desirability
Temporal
Spatial
Population Change Conceptual framework
![Page 8: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/8.jpg)
Population
Demographics
AccessibilityDevelopability
Livability Desirability
Temporal
Spatial
Population Change Conceptual frameworkLocal demographic characteristics----------------------------------------------Population densityAge: the young and the oldMinority: black and HispanicInstitutional population (college)Education attainment: HS and Bchl.Geographic mobilityPovertySeasonal housingSustenance organization: retail and agricultural industrial structure
![Page 9: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/9.jpg)
Population
Demographics
AccessibilityDevelopability
Livability Desirability
Temporal
Spatial
Population Change Conceptual framework
Transportation infrastructure--------------------------------------Residential preferenceHighway infrastructureAccessibility to airportsAccessibility to highwaysAccessibility to workplaces
![Page 10: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/10.jpg)
Population
Demographics
AccessibilityDevelopability
Livability Desirability
Temporal
Spatial
Population Change Conceptual framework
The potential for land conversion & development-----------------------------------WaterWetlandsSlopeTax-exempt (protected) landsBuilt-up lands
![Page 11: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/11.jpg)
Population
Demographics
AccessibilityDevelopability
Livability Desirability
Temporal
Spatial
Population Change Conceptual framework
Natural & built amenities desirable for living--------------------------ForestsWaterLakeshore/riverbank/ coastlineGolf coursesslope
![Page 12: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/12.jpg)
Population
Demographics
AccessibilityDevelopability
Livability Desirability
Temporal
Spatial
Population Change Conceptual framework
Urban conditions suitable for living---------------------------SafetySchool performancePublic transportationBusesPublic waterNew housingCounty seatIncomeReal estate valueEmployment rate
![Page 13: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/13.jpg)
Using Principal Components Analysis, We Developed Indices of Each of These
Conceptual Areas
Mapping the Indexes Confirmed What We know about the Areas
And the Indexes all Revealed Fairly Strong Autocorrelation
![Page 14: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/14.jpg)
Demographics
Moran’s I = 0.2878 Moran’s I = 0.4260
![Page 15: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/15.jpg)
Moran’s I = 0.4639 Moran’s I = 0.4882
Accessibility
![Page 16: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/16.jpg)
Moran’s I = 0.3565
Developability
![Page 17: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/17.jpg)
Moran’s I = 0.4089
Desirability
![Page 18: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/18.jpg)
Moran’s I = 0.7849 Moran’s I = 0.7860
Livability
![Page 19: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/19.jpg)
We Ran Lots of Regressions
Whatever the Approach, We Always Ran a Standard Normal Linear
Regression and then Corrected this Specification by Incorporating Spatial Effects (spatial lag and spatial error)
![Page 20: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/20.jpg)
OLS:
90
90
00 XP
PLn
SLM:
90
0090
90
00
P
PWLnX
P
PLn
SEM:
90
90
0090
90
00 WXP
PWLnX
P
PLn
Standard regression Spatial lag model Spatial error model Variables Coef. p-value Coef. p-value Coef. p-value Constant 0.055 0.000 0.048 0.002 0.054 0.000 Demographics 1990 0.018 0.000 0.018 0.000 0.018 0.000 Accessibility 1990 -0.014 0.000 -0.014 0.000 -0.014 0.000 Desirability 0.006 0.057 0.006 0.064 0.006 0.066 Livability 1990 0.011 0.000 0.011 0.000 0.011 0.000 Developability 0.064 0.002 0.064 0.002 0.065 0.001 Spatial parameter (λ ) / / 0.064 0.135 0.063 0.147 Measures of fit Log likelihood 899.45 900.55 901.58 AIC -1786.9 -1787.11 -1791.16
Regressions without Any Temporal Consideration
![Page 21: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/21.jpg)
OLS:
90
80
90
90
00 XP
PLn
P
PLn
SLM:
90
0090
80
90
90
00
P
PWLnX
P
PLn
P
PLn
SEM:
90
80
90
90
0090
80
90
90
00 XP
PLnW
P
PWLnX
P
PLn
P
PLn
Regressions with Temporal Consideration of Population Change
Standard regression Spatial lag model Spatial error model Variables Coef. p-value Coef. p-value Coef. p-value Constant 0.061 0.000 0.056 0.000 0.060 0.000 Population change 1980-90 0.277 0.000 0.276 0.000 0.276 0.000 Demographics 1990 0.012 0.000 0.012 0.000 0.012 0.000 Accessibility 1990 -0.011 0.000 -0.011 0.000 -0.011 0.000 Desirability 0.006 0.057 0.006 0.062 0.006 0.062 Livability 1990 0.009 0.000 0.009 0.000 0.009 0.000 Developability 0.051 0.011 0.051 0.010 0.052 0.010 Spatial parameter (λ ) / / 0.049 0.252 0.036 0.417 Measures of fit Log likelihood 942.28 942.93 943.62 AIC -1870.56 -1869.86 -1873.23
![Page 22: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/22.jpg)
Regressions with Temporal Considerations of Population Change and Indices
OLS:
8090
80
90
90
00 XXP
PLn
P
PLn
SLM:
90
008090
80
90
90
00
P
PWLnXX
P
PLn
P
PLn
SEM:
8090
80
90
90
008090
80
90
90
00 XXP
PLnW
P
PWLnXX
P
PLn
P
PLn
Standard regression Spatial lag model Spatial error model Variables Coef. p-value Coef. p-value Coef. p-value Constant 0.056 0.000 0.051 0.001 0.056 0.000 Population change 1980-90 0.275 0.000 0.274 0.000 0.273 0.000 Demographics 1990 0.004 0.387 0.004 0.397 0.004 0.408 Demographics 1980 0.008 0.067 0.008 0.065 0.008 0.062 Accessibility 1990 -0.021 0.121 -0.021 0.124 -0.021 0.123 Accessibility 1980 0.010 0.462 0.010 0.472 0.010 0.470 Desirability 0.007 0.025 0.007 0.027 0.007 0.027 Livability 1990 0.009 0.108 0.009 0.109 0.009 0.109 Livability 1980 0.001 0.873 0.001 0.871 0.001 0.845 Developability 0.057 0.005 0.057 0.005 0.058 0.005 Spatial parameter (λ ) / / 0.049 0.250 0.039 0.384 Measures of fit Log likelihood 944.21 944.86 945.60 AIC -1868.42 -1867.72 -1871.19
![Page 23: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/23.jpg)
Extrapolation projection
Baseline projection
Standard regressionPartial spatio-temporalregression
Full spatio-temporalregression
Dependent variables: population change, population density, population density changeIndices generating methods: PCA, coefficients, coefficients and correlations
Projections using indices
Population forecast adjustments
Evaluation and comparison
Projection using individual variables
Select the best one
Select the better one
Regression projection
Forecasting and Evaluation
![Page 24: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/24.jpg)
Model 1: Extrapolation projection
P P G2 0 0 0 9 0 1 0
GP P P P P P
9 0 8 0 9 0 7 0 9 0 6 0
1 0 2 0 3 03
Model 2: Standard regression
L nP
PL n
P
PX
9 0
8 0
8 0
7 08 0
L nP
PL n
P
PX
0 0
9 0
9 0
8 09 0
Model 3: partial spatio-temporal regression(incorporating spatial population effects)
L nP
PL n
P
PX W L n
P
P neighbor
90
80
80
7080
80
70
L nP
PL n
P
PX W L n
P
P neighbor
0 0
9 0
9 0
8 09 0
9 0
8 0
L nP
PL n
P
PX W L n
P
PW X
neighborneighbor
9 0
8 0
8 0
7 08 0 1
8 0
7 02 8 0
( )
L nP
PL n
P
PX W L n
P
PW X
neighborneighbor
0 0
9 0
9 0
8 09 0 1
9 0
8 02 9 0
( )
Model 4: full spatio-temporal regression(incorporating spatial population effects
and other neighbor characteristics)
Four Finalized Population Forecasting Models
![Page 25: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/25.jpg)
So… How did it turn out with all this re-engineering and fancy fuel additives?
Not well
![Page 26: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/26.jpg)
Without adjustments Model 1: extrapolation
projection
Model 2: standard
regression
Model 3: partial spatio-
temporal regression
Model 4: full spatio-
temporal regression
MPE -5.80% -4.86% -7.46% -10.05% MAPE 10.99% 10.45% 11.35% 12.86%
RMSPE 15.48% 14.56% 15.20% 16.03% MedPE -6.04% -4.76% -7.34% -8.84%
MedAPE 8.41% 7.89% 9.07% 10.06%
Population growth rate (% MCDs)
≤ -10% (5.28%) 22.71% 24.18% 20.15% 18.66% -10% < ≤ -5% (5.77%) 8.94% 9.17% 6.67% 6.41% -5% < < 0% (9.96%) 6.17% 5.77% 4.36% 4.76%
0% (0.44%) 14.46% 3.53% 2.93% 3.51% 0% < < 5% (15.41%) 5.94% 3.86% 4.34% 5.06% 5% ≤ <10% (16.28%) 7.38% 4.88% 6.75% 7.75%
≥10% (46.87%) 13.84% 14.22% 16.41% 17.82% Population size (% MCDs)
0≤ ≤ 250 (6.42%) 17.63% 15.37% 15.11% 14.98% 250< ≤ 2,000 (71.31%) 10.73% 10.08% 11.14% 11.58%
2,000< ≤ 20,000 (20.25%) 10.46% 10.78% 11.58% 13.67% >20,000 (2.02%) 4.65% 4.48% 4.66% 12.33% Metro/NonMetro (% MCDs) Metropolis, and major city
(4.68%) 6.83% 9.00% 8.44% 13.64%
Metropolis, not major city
(22.70%) 9.27% 10.73% 11.51% 12.96%
Non-Metropolis (72.62%) 9.93% 10.46% 11.49% 11.92%
Population projections to 2000 without adjustments at the MCD level
![Page 27: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/27.jpg)
With adjustments Model 1: extrapolation
projection
Model 2: standard
regression
Model 3: partial spatio-
temporal regression
Model 4: full spatio-
temporal regression
MPE -3.65% -3.79% -3.87% -3.81% MAPE 9.63% 10.69% 10.71% 10.65%
RMSPE 13.56% 14.97% 14.93% 14.84% MedPE -3.70% -4.21% -4.29% -4.25%
MedAPE 7.11% 8.19% 8.17% 8.08%
Population growth rate (% MCDs)
≤ -10% (5.28%) 23.81% 25.78% 25.36% 25.26% -10% < ≤ -5% (5.77%) 9.47% 11.13% 10.98% 11.11% -5% < < 0% (9.96%) 5.87% 7.32% 7.31% 7.24%
0% (0.44%) 4.20% 4.32% 4.70% 4.45% 0% < < 5% (15.41%) 4.50% 4.38% 4.36% 4.45% 5% ≤ <10% (16.28%) 5.23% 4.99% 5.01% 5.00%
≥10% (46.87%) 12.13% 17.36% 13.87% 13.73% Population size (% MCDs)
0≤ ≤ 250 (6.42%) 16.16% 15.25% 15.04% 14.78% 250< ≤ 2,000 (71.31%) 9.41% 10.25% 10.30% 10.25%
2,000< ≤ 2 0,000 (20.25%) 8.84% 11.33% 11.29% 11.25% >20,000 (2.02%) 4.61% 5.35% 5.53% 5.76% Metro/NonMetro (% MCDs) Metropolis, and major city
(4.68%) 6.83% 10.22% 9.86% 9.24%
Metropolis, not major city
(22.70%) 9.27% 12.09% 12.11% 11.91%
Non-Metropolis (72.62%) 9.93% 10.28% 10.33% 10.35%
Population projections to 2000 with adjustments at the MCD level
![Page 28: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/28.jpg)
Summary• Things just didn’t turn out as we hypothesized (and
hoped) they would• Our fancy spatio-temporal model outperformed
simple regression in the estimation stage of the analysis (but who cares?)
• But, to our dismay, in the forecasting stage, the a-theoretical, simple extrapolation model outperformed the regression models in all comparisons but one
• In only one set of MCDs did the fancy model outperform all others: MCDs of fewer than 250 people. We launched this project in the belief that non-demographic variables might perform best in very small areas, and this finding may suggest that we explore that possibility further
![Page 29: Test Driving a Small-Area Population Forecasting Model: Seeking Additional Horsepower through Updated Engineering and Non-Demographic Fuel Additives Paul](https://reader035.vdocuments.site/reader035/viewer/2022081516/56649ec75503460f94bd2efd/html5/thumbnails/29.jpg)
Thanks!