A Comparative Study of Machine Learning Algorithms for Industry-Specific Freight Generation Model
Abstract
:1. Introduction
2. Literature Review
- Evaluation of seven commonly used ML algorithms (i.e., Lasso, Decision Tree, Random Forest, Gradient Boosting, Support Vector, Gaussian Process, and Multi-layer Perceptron regressions), along with Ordinary Least Square (OLS) regression, with statistical tests on model performance
- Comprehensive scope of industry types—covered 45 North American Industry Classification System (NAICS) codes
- Inclusion of receipts total as an exploratory variable
- Industry-specific model selection—extensive model selection process considering model approach (ML algorithms), use of logarithm, and full combination of variable selection for each industry type
3. Data Sources
3.1. Dependent Variables—Freight Generation Data (Tonnage and Value)
- tonnage: total weight (in thousand tons) of shipments originated from (terminated to) a region by industry
- value: total value (in million dollars) of shipments originated from (terminated to) a region by industry
3.2. Independent/Explanatory Variables—Economic/Industry Data
- 2017 CBP: number of establishments (ESTAB), number of employments (EMP), and annual payroll (PAYANN)
- 2017 EC: receipt total (RCPTOT) that is the total value of sales, revenue, or shipments
3.3. Shipments by Destinations (Freight Attraction)
3.4. Descriptive Statistics of Input Data
4. Machine Learning Algorithms
4.1. Ordinary Least Squares Regression (OLS, the Baseline)
4.2. Least Absolute Shrinkage and Selection Operator (Lasso)
4.3. Decision Tree Regression (DTR)
- Maximum depth of the tree (max_depth): [1, 2, 3, 4, 5]
- Complexity parameter used for the minimal cost-complexity pruning (ccp_alpha): [0, 0.002, …, 0.018, 0.02]
4.4. Random Forest Regression (RFR)
- Maximum depth of the tree (max_depth): [1, 2, 3, 4, 5]
- Complexity parameter used for the minimal cost-complexity pruning (ccp_alpha): [0, 0.002, …, 0.018, 0.02]
- Number of trees in the forest (n_estimators): 10
4.5. Gradient Boosting Regression (GBR)
- Maximum depth of the tree (max_depth): [1, 2, 3, 4, 5]
- Complexity parameter used for the minimal cost-complexity pruning (ccp_alpha): [0, 0.002, …, 0.018, 0.02]
- Learning rate to control the contribution of each tree (learning_rate): [0.01, 0.1, 1]
- Number of the estimators (trees) (n_estimators): 10
4.6. Support Vector Regression (SVR)
- Margin of tolerance where no penalty is given to errors (epsilon): [0, 0.002, …, 0.018, 0.02]
- Regularization parameter (C): [0.1, 0.3, …, 1.9, 2.1]
- Kernel distribution type to be used in the algorithm (kernel):[Linear, Polynomial, Gaussian (RBF), Sigmoid]
4.7. Gaussian Process Regression (GPR)
- Constant value added to the diagonal of the kernel matrix (alpha): [1 × 10−11, 1 × 10−10, 1 × 10−9]
- Kernel specifying the covariance function of the model (kernel):
- Combined two kernels, Dot-Product kernel and White kernel
- For the Dot-Product kernel (DotProduct), the parameter sigma to control the inhomogeneity of the kernel: [0.5, 1.0, 1.5]
- For the White kernel (WhiteKernel), the parameter noise_level to control the noise level of the kernel: [0.5, 1.0, 1.5]
4.8. Multi-Layer Perceptron Regression (MLP)
- Hidden layer size and number of neurons in each hidden layer (hidden_layer_sizes)
- Number of hidden layers: [1, 2, 3]
- Number of neurons in each hidden layer: [3, 4, 5]
- L2 penalty parameter (alpha): [0.00001, 0.0001, 0.001]
- Activation function for the hidden layer (activation): [Identity, Logistic, Rectified Linear Unit (ReLU)]
5. Data Processing and Model Selection
5.1. Imputation of Missing Data (for CBP/EC)
5.2. Data Transformation
5.3. Normalization
5.4. Variable Selection
5.5. Optimization of Hyperparameters
5.6. Model Performance—Error Measurements
- Root Mean Square Error (RMSE): the square root of arithmetic mean of the squared difference between the 2017 CFS tonnage/value and the estimated tonnage/value
- Mean Absolute Error (MAE): the arithmetic mean of the absolute difference between the 2017 CFS tonnage/value and the estimated tonnage/value
- R-squared: the R-squared (or coefficient of determination) statistic between the 2017 CFS tonnage/value and the estimated tonnage/value
6. Model Results
6.1. Example of Model Selection—Tonnage of Shipments by Origins for NAICS 212
6.2. Significance of Model Performance Improvement by Industry
6.3. Summary of Best Model by Industry
6.4. Discussions in Model Interpretability
7. Conclusions
- Built a framework to conduct the industry-specific model selection, i.e., the variation selection, log-transform, and algorithm.
- Evaluated the significance of model improvements when using the alternative ML algorithms over the OLS for the FG modeling.
- Suggested the use of OLS regression for certain NAICSs if the RMSE reductions by the alternative ML algorithms are not statistically significant.
- Considered all possible variable combinations from the four variables in the CBP and EC data tables.
- Covered all the NAICS codes from the 2017 CFS data and estimated tonnage/value of freight shipments by both origins (generation) and destinations (attraction).
- Applying the proposed framework with use case of disaggregating freight data into more granular level of geography (e.g., county-level freight data).
- Using other external/private data sources to reveal the relationship between economy activity and associated freight shipments at individual business level.
- Expanding the model framework to forecasting future freight demand by industry type.
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Conflicts of Interest
Appendix A
NAICS Code | Description |
---|---|
212 | Mining (except oil and gas) |
311 | Food manufacturing |
312 | Beverage and tobacco product manufacturing |
313 | Textile mills |
314 | Textile product mills |
315 | Apparel manufacturing |
316 | Leather and allied product manufacturing |
321 | Wood product manufacturing |
322 | Paper manufacturing |
323 | Printing and related support activities |
324 | Petroleum and coal products manufacturing |
325 | Chemical manufacturing |
326 | Plastics and rubber products manufacturing |
327 | Nonmetallic mineral product manufacturing |
331 | Primary metal manufacturing |
332 | Fabricated metal product manufacturing |
333 | Machinery manufacturing |
334 | Computer and electronic product manufacturing |
335 | Electrical equipment, appliance, and component manufacturing |
336 | Transportation equipment manufacturing |
337 | Furniture and related product manufacturing |
339 | Miscellaneous manufacturing |
4231 | Motor vehicle and motor vehicle parts and supplies merchant wholesalers |
4232 | Furniture and home furnishing merchant wholesalers |
4233 | Lumber and other construction materials merchant wholesalers |
4234 | Professional and commercial equipment and supplies merchant wholesalers |
4235 | Metal and mineral (except petroleum) merchant wholesalers |
4236 | Household appliances and electrical and electronic goods merchant wholesalers |
4237 | Hardware, plumbing and heating equipment and supplies merchant wholesalers |
4238 | Machinery, equipment, and supplies merchant wholesalers |
4239 | Miscellaneous durable goods merchant wholesalers |
4241 | Paper and paper product merchant wholesalers |
4242 | Drugs and druggists’ sundries merchant wholesalers |
4243 | Apparel, piece goods, and notions merchant wholesalers |
4244 | Grocery and related product merchant wholesalers |
4245 | Farm product raw material merchant wholesalers |
4246 | Chemical and allied products merchant wholesalers |
4247 | Petroleum and petroleum products merchant wholesalers |
4248 | Beer, wine, and distilled alcoholic beverage merchant wholesalers |
4249 | Miscellaneous nondurable goods merchant wholesalers |
4541 | Electronic shopping and mail-order houses |
45431 | Fuel dealers |
4931 | Warehousing and storage |
5111 | Newspaper, periodical, book, and directory publishers |
551114 | Corporate, subsidiary, and regional managing offices |
References
- U.S. Department of Transportation. Bureau of Transportation Statistics and Federal Highway Administration, Freight Analysis Framework Version 5.4 (FAF5). Available online: https://www.bts.gov/faf (accessed on 25 October 2022).
- U.S. Department of Transportation. Bureau of Transportation Statistics and U.S. Department of Commerce, U.S. Census Bureau. 2017 Commodity Flow Survey. Available online: https://www2.census.gov/programs-surveys/cfs/data/2017 (accessed on 1 August 2022).
- Holguin-Veras, J.; Sarmiento, I.; Gonzalez-Calderon, C.A. Parameter Stability in Freight Generation and Distribution Demand Models in Colombia. Dyna 2011, 78, 16–20. [Google Scholar]
- Lim, R.; Qian, Z.S.; Zhang, H.M. Development of a Freight Demand Model with an Application to California. Int. J. Transp. Sci. Technol. 2014, 3, 19–38. [Google Scholar] [CrossRef] [Green Version]
- Oliveira-Neto, F.M.; Chin, S.M.; Hwang, H.L. Aggregate Freight Generation Modeling: Assessing Temporal Effect of Economic Activity on Freight Volumes with Two-Period Cross-Sectional Data. Transp. Res. Rec. 2012, 2285, 145–154. [Google Scholar] [CrossRef]
- Krisztin, T. Semi-Parametric Spatial Autoregressive Models in Freight Generation Modeling. Transp. Res. Part E Logist. Transp. Rev. 2018, 114, 121–143. [Google Scholar] [CrossRef]
- Hagenauer, J.; Helbich, M. A Comparative Study of Machine Learning Classifiers for Modeling Travel Mode Choice. Expert Syst. Appl. 2017, 78, 273–282. [Google Scholar] [CrossRef]
- Uddin, M.; Anowar, S.; Eluru, N. Modeling Freight Mode Choice Using Machine Learning Classifiers: A Comparative Study Using Commodity Flow Survey (CFS) Data. Transp. Plan. Technol. 2021, 44, 543–559. [Google Scholar] [CrossRef]
- Iranitalab, A.; Khattak, A. Comparison of Four Statistical and Machine Learning Methods for Crash Severity Prediction. Accid. Anal. Prev. 2017, 108, 27–36. [Google Scholar] [CrossRef] [PubMed]
- Rahman, S.; Bhasin, A.; Smit, A. Exploring the Use of Machine Learning to Predict Metrics Related to Asphalt Mixture Performance. Constr. Build. Mater. 2021, 295, 123585. [Google Scholar] [CrossRef]
- Salais-Fierro, T.; Martínez, A. Demand Forecasting for Freight Transport Applying Machine Learning into the Logistic Distribution. Mob. Netw. Appl. 2022, 27, 2172–2181. [Google Scholar] [CrossRef]
- Chin, S.M.; Hwang, H.L. National Freight Demand Modeling: Bridging the Gap Between Freight Flow Statistics and US Economic Patterns. In Proceedings of the 86th Annual Meeting of the Transportation Research Board, Washington, DC, USA, 21–25 January 2007. [Google Scholar]
- Novak, D.C.; Hodgdon, C.; Guo, F.; Aultman-Hall, L. Nationwide Freight Generation Models: A Spatial Regression Approach. Netw. Spat. Econ. 2011, 11, 23–41. [Google Scholar] [CrossRef]
- Bagighni, S. Volume Estimation Models for Generation and Attraction of Freight Commodity Groups Using Regression Analysis. Ph.D. Dissertation, The University of Alabama in Huntsville, Huntsville, AL, USA, 2012. [Google Scholar]
- Ha, D.H.; Combes, F. Building a Model of Freight Generation with a Commodity Flow Survey. In Commercial Transport; Springer International Publishing: Cham, Switzerland, 2016; pp. 23–37. [Google Scholar]
- Mommens, K.; Van Lier, T.; Macharis, C. Freight Demand Generation on Commodity and Loading Unit Level. Eur. J. Transp. Infrastruct. Res. 2017, 17, 1. [Google Scholar] [CrossRef]
- National Academies of Sciences, Engineering, and Medicine. NCFRP Report 37: Using Commodity Flow Survey Microdata and Other Establishment Data to Estimate the Generation of Freight, Freight Trips, and Service Trips: Guidebook; Transportation Research Board: Washington, DC, USA, 2016. [Google Scholar]
- U.S. Department of Commerce, U.S. Census Bureau. 2017 Economic Census Data. Available online: https://www.census.gov/programs-surveys/economic-census/year/2017/economic-census-2017/data.html (accessed on 1 August 2022).
- U.S. Department of Commerce, U.S. Census Bureau. 2017 County Business Patterns. Available online: https://www.census.gov/data/datasets/2017/econ/cbp/2017-cbp.html (accessed on 1 August 2022).
- Pedregosa, F.; Varoquaux, G.; Gramfort, A.; Michel, V.; Thirion, B.; Grisel, O.; Blondel, M.; Prettenhofer, P.; Weiss, R.; Dubourg, V.; et al. Scikit-learn: Machine Learning in Python. JMLR 2011, 12, 2825–2830. [Google Scholar]
Study | Study Area | Data Source | Scope of Analysis | Variables Considered | Methods Used | Model Performance |
---|---|---|---|---|---|---|
Chin and Hwang [12] | United States | Commodity Flow Survey (CFS) | CFS Area and Industry Sector | Employment and Establishment Size | OLS Regression | Except for four models, R2 > 0.70 |
Key Findings: With additional modeling efforts, the developed models could be enhanced to allow transportation analysts to assess regional economic impacts. | ||||||
Holguin-Veras et al. [3] | Colombia | Freight Origin-Destination Survey | Region (made up of municipality and 4 countries) | Gross Domestic Product (GDP), Existence of Port | OLS Regression | Adjusted R2: [0.86, 0.96] |
Key Findings: On average $1600 of GDP is needed to produce a ton of freight. | ||||||
Novak et al. [13] | United States | CFS and TranSearch | CFS Area | Population, Number of Employees, Port, Highway Length | OLS and Spatial Regression | R2: [0.33, 0.63] |
Key Findings: It is recommended to avoid the overuse and addition of highly correlated explanatory variables such as employment and population even when this improves R2; spatial regression model is the preferred specification for freight generation at the national level. | ||||||
Bagighni [14] | United States | Freight Analysis Framework (FAF) | FAF Zone and Commodity | Population, Median Age, Income, Number of Jobs by Industry Sector | OLS Regression | Adjusted R2: [0.54, 0.81] |
Key Findings: It is possible to develop good freight volume estimating models for individual commodities using regression analysis; however, the level of success for each commodity model varies. | ||||||
Oliveira-Neto et al. [5] | United States | CFS | State and Industry | Annual Payroll by Industry Sector | OLS Regression | R2: [0.40, 0.98] |
Key Findings: Payroll can explain a significant portion of the freight production at the state level for the U.S. | ||||||
Lim et al. [4] | California | FAF | FAF Zone and Commodity Group | Number of Employees, Population, Farmland Acres, Crop and Livestock Sales, Net Annual Electrical Generation using Coal | OLS Regression | R2: [0.21, 0.83] |
Key Findings: Models without constant terms have a better fit than models with constant; model fit is dependent on the commodity grouping and the choice of explanatory variables. | ||||||
Ha and Combes [15] | France | French Shipper Survey ECHO | Establishment | Employment, Economic Activity, Relations with Economic Agents, Production and Logistics Characteristics | One-way ANOVA and OLS Regression | R2: [0.16, 0.45] |
Key Findings: The number of employees and the economic sector were identified as very important explanatory variables. | ||||||
Mommens et al. [16] | Belgium | Freight volume data compiled from multiple sources | Traffic Analysis Zone and Commodity | Number of Employees, Establishment Size, Gross Floor Space, Population Density | OLS Regression | R2: [0.31, 0.69] |
Key Findings: It is doubtful that the addition of new explanatory variables will improve the model fit and consequently improvements in model accuracy. | ||||||
National Academies of Sciences, Engineering, and Medicine [17] | United States | CFS | Industry | Number of Employees | OLS Regression (linear and non-linear specifications) and Multiple Classification Analysis | Adjusted R2: [0.01, 0.73] |
Key Findings: The use of the CFS in combination with complementary datasets provides an efficient way to estimate freight generation (FG) models for the entire nation at various levels of geography; non-linear models typically provide the best representation of FG patterns. | ||||||
Krisztin [6] | European NUTS-2 regions | Eurostat | Country | Regional Share of Employment, Regional Share of Employment in Agriculture and Manufacturing, Length of Road Network, and Distance to the Closest Seaport | Spatial Autoregressive Model | Adjusted R2: [0.39, 0.87] |
Key Findings: There are significant non-linearities related to employment rates in manufacturing and infrastructure capabilities in the study regions. |
NAICS | Tonnage (Thousand Tons) | Value (Million $) | Number of Establishments (Count) | Number of Employments (Count) | Annual Payroll (Million $) | Receipt Total (Million $) | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
N | Mean | Std. 1 | N | Mean | Std. | Mean | Std. | Mean | Std. | Mean | Std. | Mean | Std. | |
212 | 119 | 23,570 | 34,993 | 118 | 710 | 1308 | 34 | 35 | 1127 | 1741 | 76 | 128 | N/A 2 | N/A |
311 | 123 | 4865 | 6957 | 127 | 6255 | 7447 | 199 | 241 | 11,466 | 12,584 | 507 | 579 | 4392 | 5726 |
312 | 107 | 1296 | 2083 | 111 | 1384 | 2948 | 62 | 110 | 1628 | 2689 | 80 | 160 | 666 | 1339 |
313 | 67 | 89 | 184 | 101 | 284 | 629 | 11 | 28 | 577 | 1604 | 24 | 64 | 121 | 393 |
314 | 79 | 54 | 195 | 116 | 214 | 656 | 35 | 46 | 687 | 1604 | 25 | 66 | 114 | 521 |
315 | 43 | 9 | 26 | 98 | 140 | 478 | 42 | 226 | 686 | 3092 | 18 | 84 | 65 | 364 |
316 | 34 | 16 | 30 | 76 | 57 | 81 | 6 | 12 | 146 | 352 | 5 | 12 | 12 | 36 |
321 | 113 | 1939 | 2876 | 125 | 889 | 1142 | 103 | 108 | 3047 | 3593 | 121 | 142 | 610 | 810 |
322 | 112 | 1411 | 1827 | 113 | 1665 | 1969 | 26 | 34 | 2103 | 2873 | 124 | 177 | 686 | 1076 |
323 | 108 | 164 | 252 | 126 | 675 | 838 | 182 | 221 | 3469 | 4140 | 155 | 204 | 535 | 751 |
324 | 98 | 12,687 | 24,707 | 112 | 4623 | 11,030 | 10 | 13 | 655 | 1363 | 68 | 154 | 2067 | 6675 |
325 | 122 | 5694 | 11,200 | 127 | 5741 | 9853 | 93 | 117 | 5618 | 7089 | 439 | 636 | 4296 | 9013 |
326 | 120 | 497 | 616 | 129 | 1869 | 2205 | 86 | 105 | 5456 | 6696 | 260 | 316 | 1391 | 1868 |
327 | 106 | 6675 | 6988 | 129 | 993 | 931 | 103 | 85 | 2897 | 2594 | 149 | 138 | 730 | 753 |
331 | 103 | 1590 | 2752 | 112 | 1915 | 2512 | 27 | 37 | 2346 | 3460 | 149 | 239 | 1011 | 1881 |
332 | 114 | 899 | 1284 | 130 | 2700 | 3016 | 402 | 463 | 10,917 | 12,494 | 558 | 663 | 2423 | 2860 |
333 | 108 | 308 | 471 | 125 | 2959 | 3377 | 169 | 200 | 7815 | 8438 | 488 | 550 | 2268 | 2774 |
334 | 79 | 31 | 46 | 118 | 2660 | 4972 | 89 | 160 | 6052 | 10,593 | 521 | 1029 | 1811 | 3814 |
335 | 94 | 160 | 222 | 115 | 1098 | 1256 | 36 | 57 | 2138 | 2938 | 135 | 208 | 503 | 881 |
336 | 97 | 1069 | 2324 | 112 | 8093 | 13,983 | 81 | 104 | 11,616 | 16,517 | 749 | 1144 | 4736 | 10,549 |
337 | 115 | 122 | 178 | 126 | 610 | 889 | 101 | 124 | 2729 | 4096 | 113 | 172 | 424 | 785 |
339 | 100 | 66 | 81 | 124 | 1274 | 1743 | 203 | 270 | 4250 | 5928 | 234 | 386 | 952 | 1652 |
4231 | 104 | 691 | 1783 | 124 | 5349 | 11,377 | 173 | 237 | 3175 | 4426 | 169 | 311 | 2287 | 6471 |
4232 | 96 | 192 | 340 | 118 | 788 | 1430 | 97 | 203 | 1357 | 2644 | 78 | 155 | 487 | 1150 |
4233 | 96 | 1703 | 2095 | 129 | 1319 | 1544 | 125 | 128 | 1912 | 2082 | 108 | 128 | 759 | 1063 |
4234 | 87 | 241 | 564 | 119 | 4183 | 8084 | 255 | 375 | 4756 | 7634 | 418 | 877 | 3123 | 7721 |
4235 | 117 | 1058 | 1669 | 124 | 1573 | 2458 | 67 | 101 | 1124 | 1714 | 70 | 117 | 737 | 2215 |
4236 | 107 | 275 | 527 | 125 | 4325 | 8658 | 208 | 349 | 3931 | 7293 | 377 | 986 | 2829 | 6764 |
4237 | 113 | 209 | 246 | 132 | 1300 | 1518 | 143 | 157 | 1974 | 2280 | 121 | 151 | 747 | 1021 |
4238 | 95 | 563 | 951 | 128 | 3879 | 4449 | 434 | 466 | 6081 | 6513 | 388 | 458 | 2589 | 3641 |
4239 | 99 | 2240 | 4068 | 125 | 1786 | 3464 | 234 | 495 | 2590 | 4369 | 136 | 245 | 440 | 1947 |
4241 | 103 | 362 | 587 | 121 | 1062 | 1819 | 67 | 115 | 1115 | 1951 | 64 | 119 | 472 | 1230 |
4242 | 86 | 196 | 782 | 112 | 6299 | 12,078 | 69 | 158 | 2188 | 5438 | 254 | 807 | 2054 | 6738 |
4243 | 86 | 102 | 382 | 103 | 1360 | 4409 | 116 | 513 | 1615 | 6106 | 98 | 403 | 935 | 4719 |
4244 | 125 | 3060 | 4220 | 131 | 6580 | 9345 | 257 | 455 | 6383 | 8787 | 338 | 484 | 3781 | 7630 |
4245 | 82 | 10,628 | 18,113 | 89 | 2314 | 3524 | 45 | 93 | 531 | 1074 | 28 | 54 | 786 | 1789 |
4246 | 110 | 1230 | 2332 | 122 | 1506 | 2792 | 84 | 113 | 1131 | 1667 | 83 | 147 | 646 | 1848 |
4247 | 116 | 11,937 | 22,794 | 122 | 6847 | 12,943 | 36 | 36 | 664 | 932 | 55 | 140 | 2986 | 24,341 |
4248 | 127 | 526 | 579 | 129 | 1262 | 1820 | 28 | 49 | 1376 | 1887 | 86 | 148 | 600 | 1519 |
4249 | 89 | 2214 | 4538 | 129 | 2410 | 3124 | 209 | 302 | 2759 | 3388 | 134 | 167 | 235 | 763 |
4541 | 99 | 207 | 433 | 123 | 4314 | 8864 | 307 | 538 | 4291 | 6698 | 194 | 353 | 3129 | 7868 |
4931 | 119 | 2235 | 2416 | 121 | 9207 | 11,193 | 120 | 144 | 6848 | 9242 | 297 | 398 | 208 | 462 |
5111 | 77 | 23 | 24 | 93 | 146 | 211 | 107 | 129 | 2661 | 4461 | 164 | 413 | 508 | 1794 |
45431 | 120 | 433 | 722 | 124 | 272 | 393 | 50 | 68 | 535 | 859 | 24 | 43 | 148 | 344 |
551114 | 56 | 362 | 475 | 69 | 1324 | 1891 | 375 | 400 | 26,406 | 34,602 | 2823 | 4615 | N/A | N/A |
NAICS | Tonnage (Thousand Tons) | Value (Million $) | Number of Establishments (Count) | Number of Employments (Count) | Annual Payroll (Million $) | Receipt Total (Million $) | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
N | Mean | Std. 1 | N | Mean | Std. | Mean | Std. | Mean | Std. | Mean | Std. | Mean | Std. | |
212 | 132 | 21,951 | 22,729 | 132 | 702 | 919 | 33 | 34 | 1090 | 1682 | 73 | 124 | 44 | 137 |
311 | 132 | 4744 | 5189 | 132 | 6041 | 7218 | 198 | 242 | 11,175 | 12,308 | 498 | 569 | 4328 | 5619 |
312 | 132 | 1188 | 1597 | 132 | 1212 | 2504 | 65 | 108 | 1899 | 2793 | 98 | 167 | 844 | 1459 |
313 | 132 | 49 | 100 | 132 | 224 | 394 | 12 | 27 | 580 | 1525 | 25 | 62 | 130 | 375 |
314 | 132 | 40 | 120 | 132 | 195 | 448 | 40 | 51 | 942 | 1617 | 39 | 68 | 181 | 486 |
315 | 132 | 4 | 6 | 132 | 106 | 198 | 37 | 208 | 606 | 2848 | 16 | 78 | 58 | 335 |
316 | 132 | 5 | 20 | 132 | 38 | 79 | 6 | 12 | 176 | 344 | 6 | 12 | 30 | 49 |
321 | 132 | 1708 | 1824 | 132 | 847 | 732 | 102 | 108 | 2998 | 3577 | 119 | 142 | 602 | 802 |
322 | 132 | 1236 | 1454 | 132 | 1454 | 1716 | 26 | 35 | 1995 | 2839 | 117 | 174 | 647 | 1053 |
323 | 132 | 154 | 237 | 132 | 650 | 819 | 171 | 207 | 3293 | 3913 | 149 | 195 | 515 | 722 |
324 | 132 | 10,438 | 21,857 | 132 | 4185 | 9811 | 10 | 13 | 620 | 1304 | 63 | 147 | 1915 | 6321 |
325 | 132 | 5412 | 9044 | 132 | 5645 | 8116 | 88 | 111 | 5305 | 6698 | 413 | 600 | 4106 | 8648 |
326 | 132 | 499 | 560 | 132 | 1826 | 2076 | 86 | 107 | 5338 | 6592 | 260 | 317 | 1433 | 1910 |
327 | 132 | 6484 | 6447 | 132 | 982 | 985 | 102 | 85 | 2899 | 2597 | 149 | 139 | 733 | 756 |
331 | 132 | 1326 | 1959 | 132 | 1721 | 2138 | 29 | 40 | 2279 | 3435 | 144 | 235 | 958 | 1823 |
332 | 132 | 864 | 1006 | 132 | 2684 | 2707 | 394 | 453 | 10,749 | 12,276 | 550 | 653 | 2401 | 2830 |
333 | 132 | 320 | 468 | 132 | 2845 | 2969 | 168 | 201 | 7697 | 8426 | 480 | 548 | 2249 | 2783 |
334 | 132 | 41 | 55 | 132 | 2437 | 4239 | 88 | 158 | 5905 | 10,349 | 504 | 1002 | 1788 | 3735 |
335 | 132 | 124 | 149 | 132 | 981 | 1059 | 37 | 58 | 2139 | 2967 | 135 | 210 | 518 | 897 |
336 | 132 | 858 | 1941 | 132 | 7276 | 12,126 | 80 | 104 | 11,087 | 16,183 | 715 | 1119 | 4506 | 10,261 |
337 | 132 | 115 | 113 | 132 | 585 | 623 | 102 | 125 | 2739 | 4075 | 114 | 172 | 435 | 785 |
339 | 132 | 58 | 73 | 132 | 1261 | 1574 | 214 | 284 | 4431 | 6095 | 250 | 404 | 1100 | 1844 |
4231 | 132 | 684 | 1202 | 132 | 5086 | 6652 | 165 | 228 | 3042 | 4260 | 162 | 299 | 2189 | 6221 |
4232 | 132 | 189 | 260 | 132 | 729 | 844 | 1626 | 2244 | 25,377 | 33,348 | 1758 | 2784 | 13,137 | 24,316 |
4234 | 132 | 208 | 308 | 132 | 3942 | 5207 | 229 | 338 | 4274 | 6882 | 376 | 790 | 2803 | 6957 |
4236 | 132 | 289 | 399 | 132 | 4168 | 5741 | 201 | 337 | 3800 | 7050 | 364 | 953 | 2735 | 6539 |
4238 | 132 | 702 | 1129 | 132 | 3791 | 3683 | 388 | 417 | 5444 | 5831 | 347 | 410 | 2317 | 3259 |
4241 | 132 | 364 | 574 | 132 | 992 | 1350 | 852 | 1571 | 16,783 | 25,333 | 1074 | 1892 | 11,693 | 32,900 |
4242 | 132 | 162 | 294 | 132 | 5964 | 7160 | 64 | 149 | 2025 | 5121 | 235 | 759 | 1900 | 6339 |
4244 | 132 | 3028 | 3764 | 132 | 6537 | 8251 | 248 | 440 | 6164 | 8486 | 326 | 468 | 3651 | 7368 |
4247 | 132 | 10,905 | 21,166 | 132 | 6467 | 12,033 | 32 | 33 | 596 | 850 | 49 | 127 | 2677 | 22,080 |
4541 | 132 | 185 | 232 | 132 | 4095 | 5009 | 353 | 559 | 4773 | 6974 | 216 | 368 | 3238 | 7905 |
4931 | 132 | 2083 | 1899 | 132 | 8955 | 8550 | 114 | 137 | 6466 | 8827 | 281 | 380 | 198 | 443 |
5111 | 132 | 31 | 64 | 132 | 181 | 354 | 79 | 96 | 1990 | 3311 | 124 | 307 | 383 | 1331 |
551114 | 132 | 540 | 1235 | 132 | 1125 | 1548 | 361 | 395 | 25,336 | 34,027 | 2696 | 4504 | N/A 2 | N/A |
Variable Selection | OLS 1 | Lasso 2 | DTR 3 | RFR 4 | GBR 5 | SVR 6 | GPR 7 | MLP 8 | |
---|---|---|---|---|---|---|---|---|---|
No Log-Transform | EMP | 26,911 | 30,419 | 32,947 | 32,672 | 30,163 | 28,093 | 26,983 | 45,787 |
PAYANN | 29,914 | 32,346 | 32,898 | 32,687 | 27,125 | 26,705 | 29,997 | 51,983 | |
ESTAB | 27,221 | 28,141 | 32,898 | 29,658 | 29,481 | 27,508 | 27,100 | 43,834 | |
EMP, PAYANN | 28,814 | 30,419 | 32,877 | 30,184 | 28,469 | 27,328 | 30,583 | 41,621 | |
EMP, ESTAB | 27,775 | 30,419 | 32,898 | 30,345 | 31,922 | 25,341 | 28,378 | 50,355 | |
PAYANN, ESTAB | 30,015 | 32,346 | 32,898 | 30,785 | 31,769 | 26,215 | 30,306 | 45,076 | |
EMP, PAYANN, ESTAB | 29,708 | 30,419 | 36,093 | 31,680 | 31,605 | 26,755 | 31,384 | 48,811 | |
Log-Transform | EMP | 25,608 | 25,742 | 29,348 | 28,221 | 30,282 | 27,645 | 25,436 | 86,158 |
PAYANN | 26,083 | 26,181 | 30,701 | 28,383 | 30,350 | 28,650 | 25,952 | 40,286 | |
ESTAB | 29,122 | 29,248 | 29,318 | 28,726 | 30,826 | 27,478 | 28,980 | 36,600 | |
EMP, PAYANN | 25,989 | 26,102 | 29,515 | 28,123 | 30,254 | 28,563 | 25,833 | 98,109 | |
EMP, ESTAB | 25,733 | 25,841 | 30,217 | 28,278 | 30,344 | 27,092 | 26,363 | 45,223 | |
PAYANN, ESTAB | 26,281 | 26,354 | 30,721 | 28,485 | 30,366 | 27,141 | 26,741 | 415,283 | |
EMP, PAYANN, ESTAB | 26,134 | 26,219 | 30,054 | 28,304 | 30,300 | 27,092 | 26,679 | 33,675 |
NAICS | Measure | Alternative | RMSE | t-Test | Wilcoxon | ||||
---|---|---|---|---|---|---|---|---|---|
OLS | Alternative | % Dif. | Stat. | p-Value | Stat. | p-Value | |||
212 | tons | SVR | 25,733 | 25,341 | −1.5% | 2.41 | 0.018 * | 1780 | 0.01 * |
value | SVR | 466 | 449 | −3.8% | 4.22 | <0.0005 * | 1566 | 0.001 * | |
311 | tons | GPR | 4447 | 3963 | −10.9% | 11.31 | <0.0005 * | 131 | <0.0005 * |
value | SVR | 3020 | 2576 | −14.7% | 8.75 | <0.0005 * | 541 | <0.0005 * | |
312 | tons | SVR | 1667 | 1414 | −15.2% | 5.79 | <0.0005 * | 891 | <0.0005 * |
value | SVR | 1985 | 1894 | −4.6% | 5.84 | <0.0005 * | 894 | <0.0005 * | |
313 | tons | OLS | 79 | - | - | - | - | - | - |
value | SVR | 239 | 226 | −5.5% | 4.80 | <0.0005 * | 1288 | <0.0005 * | |
314 | tons | GPR | 67 | 47 | −29.1% | 7.74 | <0.0005 * | 318 | <0.0005 * |
value | SVR | 158 | 144 | −9.2% | 2.78 | 0.006 * | 1413 | <0.0005 * | |
315 | tons | OLS | 11 | - | - | - | - | - | - |
value | OLS | 135 | - | - | - | - | - | - | |
316 | tons | GPR | 26 | 26 | −0.1% | 1.20 | 0.232 | 1680 | 0.004 * |
value | SVR | 68 | 66 | −2.7% | 3.57 | 0.001 * | 1713 | 0.005 * | |
321 | tons | RFR | 1946 | 1605 | −17.5% | 7.76 | <0.0005 * | 721 | <0.0005 * |
value | SVR | 384 | 311 | −19.2% | 11.19 | <0.0005 * | 247 | <0.0005 * | |
322 | tons | SVR | 1627 | 1492 | −8.3% | 12.85 | <0.0005 * | 161 | <0.0005 * |
value | SVR | 1167 | 1128 | −3.4% | 4.66 | <0.0005 * | 1148 | <0.0005 * | |
323 | tons | SVR | 179 | 150 | −16.1% | 11.12 | <0.0005 * | 183 | <0.0005 * |
value | SVR | 244 | 241 | −1.4% | 0.98 | 0.331 | 2186 | 0.244 | |
324 | tons | SVR | 11,290 | 10,730 | −5.0% | 3.86 | <0.0005 * | 1523 | 0.001 * |
value | SVR | 5009 | 4389 | −12.4% | 6.45 | <0.0005 * | 506 | <0.0005 * | |
325 | tons | GPR | 8887 | 8783 | −1.2% | 8.09 | <0.0005 * | 298 | <0.0005 * |
value | SVR | 3592 | 3409 | −5.1% | 3.36 | 0.001 * | 1945 | 0.046 * | |
326 | tons | SVR | 319 | 279 | −12.4% | 10.15 | <0.0005 * | 390 | <0.0005 * |
value | SVR | 771 | 732 | −5.1% | 7.72 | <0.0005 * | 569 | <0.0005 * | |
327 | tons | SVR | 4768 | 4441 | −6.9% | 9.39 | <0.0005 * | 383 | <0.0005 * |
value | SVR | 361 | 355 | −1.5% | 1.51 | 0.133 | 1909 | 0.034 * | |
331 | tons | SVR | 1525 | 1341 | −12.1% | 5.78 | <0.0005 * | 875 | <0.0005 * |
value | SVR | 966 | 926 | −4.1% | 7.04 | <0.0005 * | 727 | <0.0005 * | |
332 | tons | SVR | 925 | 909 | −1.7% | 1.07 | 0.286 | 2473 | 0.858 |
value | SVR | 627 | 622 | −0.8% | 1.69 | 0.095 | 1826 | 0.016 * | |
333 | tons | SVR | 333 | 231 | −30.6% | 18.83 | <0.0005 * | 9 | <0.0005 * |
value | SVR | 1391 | 1186 | −14.7% | 14.25 | <0.0005 * | 51 | <0.0005 * | |
334 | tons | SVR | 42 | 37 | −13.4% | 14.22 | <0.0005 * | - | <0.0005 * |
value | OLS | 1085 | - | - | - | - | - | - | |
335 | tons | SVR | 216 | 201 | −6.8% | 10.93 | <0.0005 * | 347 | <0.0005 * |
value | OLS | 730 | - | - | - | - | - | - | |
336 | tons | OLS | 1662 | - | - | - | - | - | - |
value | SVR | 4478 | 4437 | −0.9% | 1.16 | 0.249 | 1918 | 0.037 * | |
337 | tons | SVR | 108 | 96 | −11.2% | 10.16 | <0.0005 * | 442 | <0.0005 * |
value | OLS | 225 | - | - | - | - | - | - | |
339 | tons | DTR | 69 | 66 | −4.3% | 3.02 | 0.003 * | 1687 | 0.004 * |
value | SVR | 871 | 642 | −26.3% | 6.39 | <0.0005 * | 739 | <0.0005 * | |
4231 | tons | OLS | 876 | - | - | - | - | - | - |
value | OLS | 4492 | - | - | - | - | - | - | |
4232 | tons | SVR | 131 | 118 | −9.9% | 4.91 | <0.0005 * | 1418 | <0.0005 * |
value | SVR | 296 | 289 | −2.4% | 2.31 | 0.023 * | 2432 | 0.749 | |
4233 | tons | GBR | 1295 | 1110 | −14.3% | 6.92 | <0.0005 * | 801 | <0.0005 * |
value | SVR | 446 | 419 | −6.1% | 5.63 | <0.0005 * | 963 | <0.0005 * | |
4234 | tons | SVR | 333 | 327 | −1.7% | 1.04 | 0.299 | 1968 | 0.055 |
value | OLS | 2746 | - | - | - | - | - | - | |
4235 | tons | SVR | 969 | 878 | −9.4% | 3.47 | 0.001 * | 1520 | 0.001 * |
value | OLS | 1031 | - | - | - | - | - | - | |
4236 | tons | Lasso | 372 | 371 | −0.3% | 1.54 | 0.126 | 2312 | 0.464 |
value | SVR | 3905 | 3686 | −5.6% | 4.04 | <0.0005 * | 1291 | <0.0005 * | |
4237 | tons | OLS | 118 | - | - | - | - | - | - |
value | SVR | 543 | 513 | −5.5% | 6.31 | <0.0005 * | 924 | <0.0005 * | |
4238 | tons | SVR | 544 | 519 | −4.7% | 1.03 | 0.305 | 2048 | 0.101 |
value | Lasso | 1391 | 1382 | −0.6% | 0.92 | 0.36 | 2297 | 0.433 | |
4239 | tons | Lasso | 1494 | 1477 | −1.1% | 0.75 | 0.452 | 2356 | 0.561 |
value | Lasso | 793 | 743 | −6.4% | 1.72 | 0.089 | 2275 | 0.39 | |
4241 | tons | OLS | 382 | - | - | - | - | - | - |
value | OLS | 843 | - | - | - | - | - | - | |
4242 | tons | SVR | 574 | 568 | −1.0% | 3.33 | 0.001 * | 1642 | 0.002 * |
value | OLS | 8976 | - | - | - | - | - | - | |
4243 | tons | OLS | 96 | - | - | - | - | - | - |
value | SVR | 1175 | 845 | −28.1% | 5.69 | <0.0005 * | 971 | <0.0005 * | |
4244 | tons | OLS | 1154 | - | - | - | - | - | - |
value | OLS | 1949 | - | - | - | - | - | - | |
4245 | tons | RFR | 8771 | 8485 | −3.3% | 1.10 | 0.275 | 2367 | 0.587 |
value | DTR | 1809 | 1695 | −6.3% | 2.20 | 0.03 * | 1700 | 0.005 * | |
4246 | tons | OLS | 1525 | - | - | - | - | - | - |
value | OLS | 1075 | - | - | - | - | - | - | |
4247 | tons | OLS | 16,491 | - | - | - | - | - | - |
value | SVR | 8490 | 8245 | −2.9% | 1.78 | 0.078 | 2171 | 0.224 | |
4248 | tons | OLS | 314 | - | - | - | - | - | - |
value | SVR | 614 | 571 | −6.9% | 3.69 | <0.0005 * | 1753 | 0.008 * | |
4249 | tons | SVR | 2608 | 2235 | −14.3% | 4.93 | <0.0005 * | 1515 | 0.001 * |
value | SVR | 1685 | 1652 | −2.0% | 3.13 | 0.002 * | 1575 | 0.001 * | |
4541 | tons | SVR | 282 | 276 | −2.2% | 2.95 | 0.004 * | 1661 | 0.003 * |
value | SVR | 5102 | 4833 | −5.3% | 4.54 | <0.0005 * | 1367 | <0.0005 * | |
45431 | tons | SVR | 324 | 303 | −6.6% | 3.82 | <0.0005 * | 1520 | 0.001 * |
value | OLS | 119 | - | - | - | - | - | - | |
4931 | tons | SVR | 1494 | 1394 | −6.7% | 5.97 | <0.0005 * | 231 | <0.0005 * |
value | SVR | 5854 | 5505 | −6.0% | 3.98 | <0.0005 * | 794 | <0.0005 * | |
5111 | tons | SVR | 21 | 20 | −6.3% | 8.79 | <0.0005 * | 493 | <0.0005 * |
value | SVR | 183 | 182 | −0.4% | 0.94 | 0.352 | 1852 | 0.021 * | |
551114 | tons | RFR | 493 | 480 | −2.7% | 2.39 | 0.019 * | 1680 | 0.004 * |
value | DTR | 1733 | 1653 | −4.7% | 3.80 | <0.0005 * | 1503 | <0.0005 * | |
(* p-value < 0.05) |
NAICS | Measure | Alternative | RMSE | t-Test | Wilcoxon | ||||
---|---|---|---|---|---|---|---|---|---|
OLS | Alternative | %Dif. | Stat. | p-Value | Stat. | p-Value | |||
212 | tons | RFR | 18,452 | 17,962 | −2.7% | 2.67 | 0.009 * | 1795 | 0.012 * |
value | SVR | 770 | 756 | −1.8% | 5.20 | <0.0005 * | 1042 | <0.0005 * | |
311 | tons | GPR | 2083 | 2078 | −0.3% | 0.54 | 0.59 | 2095 | 0.139 |
value | SVR | 3537 | 3482 | −1.6% | 1.87 | 0.065 | 1843 | 0.019 * | |
312 | tons | SVR | 1077 | 1050 | −2.6% | 4.86 | <0.0005 * | 1038 | <0.0005 * |
value | SVR | 1827 | 1796 | −1.7% | 4.08 | <0.0005 * | 1538 | 0.001 * | |
313 | tons | GPR | 73 | 73 | −0.3% | 0.47 | 0.638 | 1643 | 0.002 * |
value | SVR | 261 | 254 | −2.9% | 4.28 | <0.0005 * | 1332 | <0.0005 * | |
314 | tons | GPR | 47 | 43 | −7.6% | 6.17 | <0.0005 * | 776 | <0.0005 * |
value | GPR | 121 | 120 | −0.2% | 2.13 | 0.036 * | 2036 | 0.093 | |
315 | tons | GPR | 4 | 4 | −3.8% | 3.82 | <0.0005 * | 1428 | <0.0005 * |
value | GPR | 106 | 91 | −13.8% | 7.13 | <0.0005 * | 728 | <0.0005 * | |
316 | tons | MLP | 15 | 14 | −6.6% | 1.32 | 0.189 | 2513 | 0.967 |
value | SVR | 61 | 49 | −20.2% | 7.25 | <0.0005 * | 657 | <0.0005 * | |
321 | tons | SVR | 1154 | 961 | −16.8% | 18.37 | <0.0005 * | 14 | <0.0005 * |
value | RFR | 443 | 419 | −5.4% | 4.67 | <0.0005 * | 1269 | <0.0005 * | |
322 | tons | SVR | 709 | 693 | −2.2% | 3.28 | 0.001 * | 1883 | 0.027 * |
value | GPR | 691 | 691 | −0.1% | 1.09 | 0.278 | 2263 | 0.368 | |
323 | tons | GPR | 135 | 133 | −2.0% | 5.03 | <0.0005 * | 852 | <0.0005 * |
value | SVR | 253 | 244 | −3.6% | 1.58 | 0.118 | 2243 | 0.332 | |
324 | tons | GPR | 10,031 | 10,017 | −0.1% | 1.38 | 0.17 | 2305 | 0.449 |
value | GPR | 4476 | 3775 | −15.7% | 11.84 | <0.0005 * | 175 | <0.0005 * | |
325 | tons | GPR | 5442 | 5437 | −0.1% | 1.22 | 0.226 | 2194 | 0.255 |
value | GPR | 3960 | 3957 | −0.1% | 2.34 | 0.021 * | 1952 | 0.049 * | |
326 | tons | SVR | 260 | 257 | −1.0% | 1.82 | 0.072 | 2016 | 0.08 |
value | SVR | 868 | 864 | −0.4% | 0.81 | 0.418 | 2321 | 0.483 | |
327 | tons | DTR | 4436 | 4119 | −7.1% | 4.22 | <0.0005 * | 1347 | <0.0005 * |
value | SVR | 364 | 355 | −2.4% | 4.55 | <0.0005 * | 1037 | <0.0005 * | |
331 | tons | GPR | 1276 | 1244 | −2.5% | 7.74 | <0.0005 * | 701 | <0.0005 * |
value | GPR | 1298 | 1270 | −2.2% | 7.75 | <0.0005 * | 701 | <0.0005 * | |
332 | tons | SVR | 692 | 615 | −11.1% | 5.62 | <0.0005 * | 989 | <0.0005 * |
value | SVR | 1150 | 1114 | −3.1% | 2.01 | 0.047 * | 2323 | 0.487 | |
333 | tons | SVR | 353 | 284 | −19.6% | 9.52 | <0.0005 * | 159 | <0.0005 * |
value | SVR | 1801 | 1782 | −1.0% | 2.17 | 0.032 * | 1860 | 0.022 * | |
334 | tons | GPR | 43 | 42 | −3.3% | 6.20 | <0.0005 * | 889 | <0.0005 * |
value | OLS | 2181 | - | - | - | - | - | - | |
335 | tons | GBR | 132 | 127 | −3.6% | 1.94 | 0.055 | 1910 | 0.034 * |
value | SVR | 656 | 639 | −2.6% | 3.24 | 0.002 * | 2418 | 0.713 | |
336 | tons | GPR | 1483 | 1049 | −29.3% | 6.61 | <0.0005 * | 1018 | <0.0005 * |
value | SVR | 5910 | 5813 | −1.6% | 0.87 | 0.387 | 2143 | 0.189 | |
337 | tons | GPR | 61 | 61 | −0.3% | 0.89 | 0.376 | 2508 | 0.953 |
value | SVR | 304 | 269 | −11.6% | 10.23 | <0.0005 * | 261 | <0.0005 * | |
339 | tons | GPR | 39 | 38 | −2.4% | 6.02 | <0.0005 * | 856 | <0.0005 * |
value | RFR | 853 | 717 | −16.0% | 7.52 | <0.0005 * | 696 | <0.0005 * | |
4231 | tons | SVR | 612 | 576 | −5.9% | 4.14 | <0.0005 * | 1496 | <0.0005 * |
value | Lasso | 2019 | 2018 | 0.0% | 2.16 | 0.033 * | 2233 | 0.315 | |
4232 | tons | OLS | 178 | - | - | - | - | - | - |
value | SVR | 371 | 335 | −9.8% | 3.45 | 0.001 * | 2323 | 0.487 | |
4234 | tons | GPR | 168 | 165 | −1.6% | 2.28 | 0.025 * | 1476 | <0.0005 * |
value | SVR | 1765 | 1697 | −3.9% | 3.98 | <0.0005 * | 1463 | <0.0005 * | |
4236 | tons | GPR | 264 | 261 | −1.4% | 3.95 | <0.0005 * | 1536 | 0.001 * |
value | OLS | 2458 | - | - | - | - | - | - | |
4238 | tons | SVR | 861 | 849 | −1.4% | 2.34 | 0.021 * | 1502 | <0.0005 * |
value | SVR | 1892 | 1817 | −4.0% | 4.53 | <0.0005 * | 1040 | <0.0005 * | |
4241 | tons | GPR | 311 | 306 | −1.9% | 2.84 | 0.005 * | 1502 | <0.0005 * |
value | GPR | 580 | 560 | −3.5% | 6.63 | <0.0005 * | 646 | <0.0005 * | |
4242 | tons | SVR | 218 | 212 | −3.1% | 3.50 | 0.001 * | 1819 | 0.015 * |
value | SVR | 4406 | 3845 | −12.7% | 6.13 | <0.0005 * | 940 | <0.0005 * | |
4244 | tons | GPR | 1151 | 1056 | −8.3% | 7.66 | <0.0005 * | 513 | <0.0005 * |
value | GPR | 1737 | 1639 | −5.6% | 6.49 | <0.0005 * | 628 | <0.0005 * | |
4247 | tons | GPR | 15,045 | 14,784 | −1.7% | 2.26 | 0.026 * | 1594 | 0.001 * |
value | GPR | 7666 | 7344 | −4.2% | 4.15 | <0.0005 * | 1416 | <0.0005 * | |
4541 | tons | SVR | 131 | 121 | −7.4% | 5.39 | <0.0005 * | 1185 | <0.0005 * |
value | SVR | 1596 | 1552 | −2.8% | 2.11 | 0.037 * | 2210 | 0.279 | |
4931 | tons | GPR | 1131 | 1014 | −10.3% | 6.12 | <0.0005 * | 863 | <0.0005 * |
value | GPR | 4358 | 3964 | −9.0% | 8.70 | <0.0005 * | 586 | <0.0005 * | |
5111 | tons | GBR | 43 | 40 | −6.5% | 1.98 | 0.051 | 2314 | 0.468 |
value | GPR | 204 | 201 | −1.8% | 4.63 | <0.0005 * | 1871 | 0.025 * | |
551114 | tons | RFR | 1012 | 984 | −2.8% | 3.93 | <0.0005 * | 1302 | <0.0005 * |
value | SVR | 1194 | 1157 | −3.1% | 7.04 | <0.0005 * | 609 | <0.0005 * | |
(* p-value < 0.05) |
NAICS | Measure | Shipments by Origins (Freight Production) | Shipments by Destinations (Freight Attraction) | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Model | Log | ESTAB | EMP | PAYANN | RCPTOT | Model | Log | ESTAB | EMP | PAYANN | RCPTOT | ||
212 | tons | SVR | No | √ | √ | RFR | Yes | √ | √ | ||||
value | SVR | No | √ | √ | √ | SVR | Yes | √ | √ | ||||
311 | tons | GPR | Yes | √ | √ | √ | GPR | No | √ | √ | |||
value | SVR | Yes | √ | √ | √ | SVR | No | √ | √ | √ | √ | ||
312 | tons | SVR | No | √ | √ | SVR | Yes | √ | |||||
value | SVR | Yes | √ | √ | √ | √ | SVR | Yes | √ | √ | √ | ||
313 | tons | OLS | No | √ | GPR | Yes | √ | ||||||
value | SVR | No | √ | √ | SVR | Yes | √ | ||||||
314 | tons | GPR | No | √ | √ | √ | GPR | No | √ | √ | |||
value | SVR | No | √ | √ | √ | GPR | No | √ | |||||
315 | tons | OLS | No | √ | GPR | Yes | √ | √ | |||||
value | OLS | No | √ | GPR | Yes | √ | √ | √ | √ | ||||
316 | tons | GPR | Yes | √ | MLP | Yes | √ | √ | |||||
value | SVR | No | √ | √ | SVR | No | √ | √ | √ | ||||
321 | tons | RFR | No | √ | SVR | Yes | √ | √ | |||||
value | SVR | Yes | √ | √ | RFR | Yes | √ | √ | |||||
322 | tons | SVR | Yes | √ | √ | SVR | No | √ | √ | ||||
value | SVR | No | √ | √ | √ | √ | GPR | No | √ | √ | |||
323 | tons | SVR | Yes | √ | √ | GPR | Yes | √ | √ | ||||
value | SVR | Yes | √ | √ | √ | SVR | Yes | √ | √ | ||||
324 | tons | SVR | No | √ | √ | √ | √ | GPR | No | √ | |||
value | SVR | No | √ | √ | √ | √ | GPR | No | √ | √ | |||
325 | tons | GPR | Yes | √ | GPR | No | √ | ||||||
value | SVR | No | √ | √ | GPR | No | √ | ||||||
326 | tons | SVR | No | √ | √ | √ | SVR | No | √ | √ | √ | √ | |
value | SVR | No | √ | √ | SVR | No | √ | √ | √ | ||||
327 | tons | SVR | No | √ | DTR | Yes | √ | √ | |||||
value | SVR | Yes | √ | √ | SVR | No | √ | √ | √ | √ | |||
331 | tons | SVR | Yes | √ | √ | √ | GPR | Yes | √ | √ | |||
value | SVR | No | √ | √ | √ | √ | GPR | Yes | √ | √ | √ | ||
332 | tons | SVR | No | √ | √ | SVR | No | √ | √ | ||||
value | SVR | Yes | √ | SVR | No | √ | √ | ||||||
333 | tons | SVR | No | √ | √ | √ | SVR | No | √ | √ | |||
value | SVR | No | √ | √ | √ | SVR | No | √ | √ | ||||
334 | tons | SVR | No | √ | GPR | No | √ | √ | √ | ||||
value | OLS | No | √ | OLS | No | √ | |||||||
335 | tons | SVR | Yes | √ | √ | √ | GBR | Yes | √ | √ | |||
value | OLS | No | √ | √ | SVR | No | √ | ||||||
336 | tons | OLS | Yes | √ | √ | GPR | No | √ | √ | √ | |||
value | SVR | No | √ | √ | √ | √ | SVR | No | √ | √ | √ | ||
337 | tons | SVR | No | √ | √ | √ | GPR | Yes | √ | √ | |||
value | OLS | No | √ | √ | √ | SVR | Yes | √ | √ | √ | |||
339 | tons | DTR | Yes | √ | √ | GPR | No | √ | √ | ||||
value | SVR | No | √ | √ | √ | RFR | Yes | √ | √ | ||||
4231 | tons | OLS | Yes | √ | SVR | Yes | √ | √ | √ | √ | |||
value | OLS | Yes | √ | √ | Lasso | No | √ | √ | |||||
4232 | tons | SVR | No | √ | √ | √ | OLS | Yes | √ | √ | |||
value | SVR | No | √ | √ | SVR | No | √ | √ | |||||
4233 | tons | GBR | No | √ | √ | √ | N/A | ||||||
value | SVR | No | √ | √ | √ | ||||||||
4234 | tons | SVR | Yes | √ | √ | GPR | Yes | √ | √ | √ | |||
value | OLS | Yes | √ | √ | SVR | No | √ | √ | √ | ||||
4235 | tons | SVR | No | √ | √ | N/A | |||||||
value | OLS | No | √ | ||||||||||
4236 | tons | Lasso | Yes | √ | GPR | Yes | √ | √ | |||||
value | SVR | No | √ | √ | √ | √ | OLS | Yes | √ | ||||
4237 | tons | OLS | Yes | √ | √ | N/A | |||||||
value | SVR | No | √ | √ | √ | ||||||||
4238 | tons | SVR | Yes | √ | √ | SVR | Yes | √ | √ | √ | |||
value | Lasso | Yes | √ | SVR | Yes | √ | √ | √ | |||||
4239 | tons | Lasso | Yes | √ | √ | N/A | |||||||
value | Lasso | No | √ | √ | |||||||||
4241 | tons | OLS | Yes | √ | GPR | Yes | √ | √ | |||||
value | OLS | Yes | √ | √ | GPR | Yes | √ | √ | |||||
4242 | tons | SVR | No | √ | SVR | Yes | √ | √ | |||||
value | OLS | Yes | √ | √ | SVR | Yes | √ | √ | √ | ||||
4243 | tons | OLS | No | √ | √ | N/A | |||||||
value | SVR | No | √ | √ | |||||||||
4244 | tons | OLS | No | √ | GPR | Yes | √ | √ | |||||
value | OLS | No | √ | √ | GPR | Yes | √ | √ | |||||
4245 | tons | RFR | Yes | √ | √ | N/A | |||||||
value | DTR | No | √ | √ | |||||||||
4246 | tons | OLS | Yes | √ | N/A | ||||||||
value | OLS | No | √ | ||||||||||
4247 | tons | OLS | Yes | √ | GPR | Yes | √ | √ | |||||
value | SVR | Yes | √ | √ | GPR | Yes | √ | √ | |||||
4248 | tons | OLS | No | √ | N/A | ||||||||
value | SVR | No | √ | √ | √ | ||||||||
4249 | tons | SVR | Yes | √ | √ | √ | √ | N/A | |||||
value | SVR | Yes | √ | √ | |||||||||
4541 | tons | SVR | Yes | √ | √ | SVR | No | √ | √ | ||||
value | SVR | Yes | √ | √ | √ | SVR | No | √ | √ | √ | |||
4931 | tons | SVR | Yes | √ | GPR | No | √ | √ | |||||
value | SVR | Yes | √ | GPR | No | √ | √ | ||||||
5111 | tons | SVR | Yes | √ | √ | GBR | Yes | √ | |||||
value | SVR | Yes | √ | √ | GPR | Yes | √ | √ | √ | ||||
45431 | tons | SVR | No | √ | √ | N/A | |||||||
value | OLS | No | √ | ||||||||||
551114 | tons | RFR | Yes | √ | RFR | Yes | √ | ||||||
value | DTR | Yes | √ | SVR | Yes | √ | √ | ||||||
(√: the variable is included in the final model) |
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Lim, H.; Uddin, M.; Liu, Y.; Chin, S.-M.; Hwang, H.-L. A Comparative Study of Machine Learning Algorithms for Industry-Specific Freight Generation Model. Sustainability 2022, 14, 15367. https://doi.org/10.3390/su142215367
Lim H, Uddin M, Liu Y, Chin S-M, Hwang H-L. A Comparative Study of Machine Learning Algorithms for Industry-Specific Freight Generation Model. Sustainability. 2022; 14(22):15367. https://doi.org/10.3390/su142215367
Chicago/Turabian StyleLim, Hyeonsup, Majbah Uddin, Yuandong Liu, Shih-Miao Chin, and Ho-Ling Hwang. 2022. "A Comparative Study of Machine Learning Algorithms for Industry-Specific Freight Generation Model" Sustainability 14, no. 22: 15367. https://doi.org/10.3390/su142215367