turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- SAS Programming
- /
- General Programming
- /
- Proc Logistic - Hosmer and Lemeshow Goodness-of-Fi...

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

01-02-2014 10:08 AM

Hi,

anybudy can guide me if my Hosmer and Lemeshow Goodness-of-Fit output is as bellow.... how it will have impact on model?

Hosmer and Lemeshow Goodness-of-Fit | ||

Test | ||

Chi-Square | DF | Pr > ChiSq |

363.6838 | 8 | <.0001 |

**Othe SAS output is good as below only Hosmer and Lemeshow Goodness-of-Fit is not good:**

Association of Predicted Probabilities and Observed Responses | |||

Percent Concordant | 85.2 | Somers' D | 0.705 |

Percent Discordant | 14.7 | Gamma | 0.706 |

Percent Tied | 0.1 | Tau-a | 0.33 |

Pairs | 1.37E+09 | c | 0.853 |

Testing Global Null Hypothesis: BETA=0 | |||

Test | Chi-Square | DF | Pr > ChiSq |

Likelihood Ratio | 29145.53 | 12 | <.0001 |

Score | 14436.35 | 12 | <.0001 |

Wald | 18640.18 | 12 | <.0001 |

Fit Statistics for SCORE Data | |||||||||||

Data Set | Total Frequency | Log Likelihood | Error Rate | AIC | AICC | BIC | SC | R-Square | Max-Rescaled R-Square | AUC | Brier Score |

WORK.SAMPLE | 76553 | -36063 | 0.22 | 72152.01 | 72152.01 | 72272.2 | 72272.2 | 0.316633 | 0.431594 | 0.852573 | 0.150807 |

WORK.HOLD_OUT | 32802 | -15594 | 0.2217 | 31213.95 | 31213.96 | 31323.13 | 31323.13 | 0.31073 | 0.423544 | 0.852483 | 0.151017 |

Thanks,

KP

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Kpatel306

01-02-2014 12:20 PM

HL test in proc logistic uses 10 groups to measure the goodness of fit, i.e, the kai square test always has 8 degrees of freedom. However, based on the predictors, the number of natural groups is usually not 10, i.e. the test statistics doesn't follow the distribution of kai square (8) .

I don't look at the statistics usually. Your C value is 0.85, that is pretty good. I think you got a good model.