MACHINE LEARNING FOR FINANCE - BLANK MODEL SET: THE BOOK'S OWN INPUTS

The four workbooks share one Inputs sheet. Type these 15 values into it - in any of the four files - and
the book's results come back. Cell references are on the Inputs sheet. The 8,400 applications on the
Data sheet are not inputs and stay in every file.

   Inputs!B7    Points to double the odds (PDO)                            20
   Inputs!B8    Anchor score                                               600
   Inputs!B9    Anchor odds (good to bad)                                  50
   Inputs!B16   Average exposure                                           185000
   Inputs!B17   Lifetime net interest margin                               0.0412
   Inputs!B18   Loss given default                                         0.417
   Inputs!B19   Cost of underwriting one case                              640
   Inputs!B25   Intercept                                                  -2.698320182594872
   Inputs!B26   Age of business (months)                                   -1.002075798703775
   Inputs!B27   Debt service cover                                         -1.088979548830445
   Inputs!B28   Net debt / EBITDA                                          -0.9933795860933711
   Inputs!B29   Sector                                                     -1.080968398212232
   Inputs!B30   Worst CCJ                                                  -1.035122757690365
   Inputs!B31   Bank conduct score                                         -1.147511872908931
   Inputs!B32   Loan to value                                              -1.168635723925453

Typed back in, they give 28.8539 and 487.1229 for the factor and offset, an area under the curve
of 0.735329, a stability index of 0.090517, a break-even bad rate of 8.2367 per cent, and 17,348,284
at a cut-off of 550 on the development book.
