Num paid Number of past loans paid off in full Loan type Either a first O for

Num paid number of past loans paid off in full loan

This preview shows page 14 - 16 out of 16 pages.

Num paid Number of past loans paid off in full. Loan type Either a first [O for original] loan made to this merchant or a repeat [R] loan. Region US time zones: East, Central, Mountain, West and other (Alaska, Hawaii) Summary of Fit, Response = PRSM RSquare 0.501216 Root Mean Square Error 0.053019 Mean of Response 0.866765 Observations (or Sum Wgts) 990 Analysis of Variance Source DF Sum of Squares Mean Square F Ratio Model 15 2.7512911 0.183419 65.2499 Error 974 2.7379410 0.002811 Prob > F C. Total 989 5.4892321 <.0001* Effect Tests Source DF Sum of Squares F Ratio Prob > F Stress 1 0.01305124 4.6429 0.0314* log (Home value) 1 0.11835042 42.1022 <.0001* Commission ratio 1 0.03678939 13.0875 0.0003* Residential ratio 1 0.40613873 144.4805 <.0001* log(1+Num Paid) 1 0.55455095 197.2769 <.0001* Loan.Type 1 0.00168410 0.5991 0.4391 Region 4 0.17822840 15.8508 <.0001* Stress*Region 4 0.07534193 6.7006 <.0001* FICO 1 0.50800297 180.7179 <.0001*
Background image
Statistics 621, Final Exam -15 of 16- Fall, 2010 Indicator Function Parameterization Term Estimate Std Error t Ratio Prob>|t| Intercept 0.1760 0.0541 3.26 0.0012* FICO 0.0003 0.0000 13.44 <.0001* Stress -0.0558 0.0100 -5.60 <.0001* log (Home value) 0.0231 0.0036 6.49 <.0001* Commission ratio -0.1501 0.0415 -3.62 0.0003* Residential ratio 0.2852 0.0237 12.02 <.0001* log(1+Num Paid) 0.0308 0.0022 14.05 <.0001* Loan type[O] -0.0031 0.0040 -0.77 0.4391 Region[Central] -0.0202 0.0185 -1.09 0.2748 Region[East] -0.0068 0.0162 -0.42 0.6737 Region[Mountain] -0.1835 0.0256 -7.16 <.0001* Region[Other] -0.0953 0.0610 -1.56 0.1188 Stress*Region[Central] 0.0079 0.0126 0.63 0.5278 Stress*Region[East] 0.0047 0.0111 0.42 0.6724 Stress*Region[Mountain] 0.0866 0.0187 4.62 <.0001* Stress*Region[Other] 0.0742 0.0430 1.73 0.0843 38.The addition of Stress*Regionto this regression is statistically significant because a)The p-value of its effect test is less than 0.05. b)One of the p-values for its terms has a * by the value. c)The t-ratio for one of its coefficients is larger than 2 in absolute size. d)The addition of these terms increased the r-squared statistic of the model. e)Overall F-ratio of the model is larger than 4. 39.A business is located in an area with homes valued at $150,000 on average. When evaluating the loan, a data error valued the homes at $160,000. This error 40.The fitted equation of this model shows that increases in stress, assuming other characteristics of the borrower remain fixed, have the largest effect on the predicted values (in magnitude) in the 41.Consider making a new loan to a business that has paid off 3 loans, has FICO score 600, stress level 0.3, is located in an area which is 90% residential (Residential ratio 0.9) with homes that average $150,000, and has commission ratio 0.15. The largest predicted PRSM is for a business in the
Background image
Statistics 621, Final Exam -16 of 16- Fall, 2010
Background image

You've reached the end of your free preview.

Want to read all 16 pages?

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture