为什么glmulti R包中的函数在lmer fit(线性混合模型)和gls拟合模型(lme包)上不能很好地工作:
一个。提取模型平均系数?它的coef函数不起作用。
我在lmer拟合模型(即混合模型)上使用glmulti R包并运行模型选择。但是我没有进行模型平均,因为即使我应用了这里提到的包装器getfit()函数,{@ 3}}
也没有使用coef函数。B中。它的级别= 2选择,即专用于包括glmulti对象上的成对交互的部分? 它有时会起作用,而在另一种情况下会失败。例如,我在失败时收到此错误消息。我选择了不同的方法=“h”,“g”,“d”来看看失败是否与计算能力有关,但没有一个选择有效。 “.jnew中的错误(”glmulti / ModelGenerator“,y,.jarray(xc),. jarray(xq),: java.lang.ArrayIndexOutOfBoundsException:10“ 这个问题的另一个问题是,一旦我在新的glmulti上收到此错误消息,那些运行良好的先前作品将无法再次运行。
℃。如果我使用MuMIn包中的模型平均函数并根据其输出进行推理,会产生多大的差异?我关注的是MuMIn包中glmulti包的作者给出的批评。他们说“* MuMIn可以处理包含交互的公式,但它将交互视为标准变量,这引发了一些问题”,请参阅第4页第二段glmulti and liner mixed models。
非常感谢你的帮助:)。
感谢Ben的快速回复和建议。这是我的数据。块和组合用作随机效应因子,六个变量(TShann,Alt,Slope,CPT,MAT和MARF)用作固定效应因子(协变量)。我想调查这六个变量对Yield的主要和成对交互作用。
Blocks TShann Alt Slope CPT MAT MARF PlotID Layer Composition Yeild
Block1 1.82 87 1 98.65 2.6 625 B1P1 0-10cm Pa,Ps 37.42
Block1 1.77 138 1 25.71 2.4 638 B1P2 0-10cm Bp,Pa 42.47
Block1 1.57 139 1 16.5 2.4 638 B1P3 0-10cm Bp,Pa 54.87
Block1 1.93 138 1 63.3 2.5 637 B1P4 0-10cm Bp,Pa 51.93
Block1 1.89 114 2 75.11 2.6 631 B1P5 0-10cm Bp,Ps 27.27
Block1 1.04 112 1 99.39 2.5 631 B1P6 0-10cm Pa 47.66
Block1 1.02 120 1 0.31 2.3 625 B1P7 0-10cm Bp 47.62
Block1 1.06 120 1 0.98 2.3 624 B1P8 0-10cm Bp 41.31
Block1 1.09 119 1 99.08 2.2 623 B1P9 0-10cm Ps 39.69
Block1 1.07 134 1 98.77 2.1 624 B1P10 0-10cm Pa 46.55
Block1 1.12 124 2 2.48 2.2 623 B1P11 0-10cm Bp 40.55
Block1 2.45 233 1 74.2 1.4 639 B1P12 0-10cm Bp,Pa,Ps 40.28
Block1 2 219 2 79.15 1.4 639 B1P13 0-10cm Bp,Ps 25.31
Block1 1 101 1 100 1.8 622 B1P14 0-10cm Ps 22.72
Block1 1.8 97 1 76.35 1.9 622 B1P15 0-10cm Bp,Ps 28.54
Block1 1.16 143 1 97.95 1.8 634 B1P16 0-10cm Ps 22.4
Block1 1.73 139 2 100 1.8 633 B1P17 0-10cm Pa,Ps 24.26
Block1 1.05 110 1 99.19 2 633 B1P18 0-10cm Pa 33.76
Block1 2.03 130 2 99.75 2.1 634 B1P19 0-10cm Pa,Ps 36.86
Block1 1.57 119 1 83.18 2.2 628 B1P20 0-10cm Bp,Pa 42.5
Block1 1.05 135 2 100 2 637 B1P21 0-10cm Ps 22.44
Block1 1.99 126 1 100 2.1 635 B1P22 0-10cm Pa,Ps 39.58
Block1 1.18 122 1 100 2.1 634 B1P23 0-10cm Pa 37.67
Block1 1.9 151 2 74 1.9 637 B1P24 0-10cm Bp,Pa 49.77
Block1 2.55 136 1 86.87 2.1 635 B1P25 0-10cm Bp,Pa,Ps 38.21
Block1 2.97 108 1 70.06 2.1 636 B1P26 0-10cm Bp,Pa,Ps 31.7
Block1 2.31 119 1 91.18 2.5 636 B1P27 0-10cm Pa,Ps 36.8
Block1 2.13 111 2 51.16 2.5 636 B1P28 0-10cm Bp,Pa 45.83
Block1 1 422.8 2 0 7.3 711 B2P1 0-10cm Fs 37.59
Block2 1 389.93 1 0 7.4 697 B2P2 0-10cm Fs 35.08
Block2 1.68 323.1 1 2.96 7.8 636 B2P3 0-10cm Ap,Fe 40.07
Block2 1.43 272.7 1 0 7.9 631 B2P4 0-10cm Fe 38.47
Block2 1.19 337.04 1 0 7.8 637 B2P5 0-10cm Qp 34.49
Block2 2 284 1 0 7.8 638 B2P6 0-10cm Fs,Qp 30.55
Block2 1.3 479.9 1 92.69 6.9 755 B2P7 0-10cm Pa 60.06
Block2 1.27 328.52 1 94.19 7.8 643 B2P8 0-10cm Pa 35.41
Block2 2.86 371.96 1 0 7.4 697 B2P9 0-10cm Ap,Fe,Fs 33.21
Block2 2.94 381.31 2 0 7.4 697 B2P10 0-10cm Ap,Fe,Fs 41.88
Block2 2.36 457.82 1 0 7.1 736 B2P11 0-10cm Fe,Fs 39.86
Block2 2.29 382.73 1 0 7.3 694 B2P12 0-10cm Fs,Qp 32.56
Block2 2.8 301.09 1 0 8 611 B2P13 0-10cm Ap,Fe,Qp 33.91
Block2 2.49 323.66 1 63.24 7.8 643 B2P14 0-10cm Fs,Pa 39.31
Block2 1.84 378.12 2 69.99 7.4 699 B2P15 0-10cm Fe,Pa 59.51
Block2 2.72 331.15 2 47.28 7.6 661 B2P16 0-10cm Fs,Pa,Qp 35.54
Block2 1.68 364.72 1 0 7.5 678 B2P17 0-10cm Ap,Fs 31.34
Block2 2.3 494.38 1 1.86 6.9 765 B2P18 0-10cm Ap,Fe 42.31
Block2 2.17 407.45 1 0 7.3 694 B2P19 0-10cm Ap,Fs,Qp 34.71
Block2 2.79 324.78 1 0 7.7 637 B2P20 0-10cm Fe,Fs,Qp 38.69
Block2 2.91 391.97 1 0 7.3 694 B2P21 0-10cm Ap,Fe,Fs,Qp 37.87
Block2 2.52 382.84 1 4.5 7.4 691 B2P22 0-10cm Fe,Fs 31.91
Block2 2.78 378.29 1 26.12 7.5 658 B2P23 0-10cm Fs,Pa,Qp 36.13
Block2 2.84 418.17 2 0 7.2 737 B2P24 0-10cm Ap,Fe,Fs 37.06
Block2 2.86 368.38 1 0 7.5 671 B2P25 0-10cm Ap,Fs,Qp 30.2
Block2 3.03 391.77 2 4.01 7 755 B2P26 0-10cm Ap,Fe,Fs 81.08
Block2 3.38 469.13 1 0 7 746 B2P27 0-10cm Ap,Fe,Fs,Qp 43.39
Block2 2 282.63 1 0 8 611 B2P28 0-10cm Ap,Fe 55.47
Block2 3.15 401.12 2 3.27 7.3 699 B2P29 0-10cm Fe,Fs,Qp 43.54
Block2 2.31 415.49 3 0 7.2 737 B2P30 0-10cm Fs,Qp 36.96
Block2 2.87 458.57 1 0 6.9 758 B2P31 0-10cm Fe,Fs 43.75
Block2 3.54 387.87 1 0 7.1 745 B2P32 0-10cm Ap,Fe,Fs 38.18
Block2 3.81 390.85 1 15.3 7.5 681 B2P33 0-10cm Ap,Fe,Fs,Pa 56.03
Block2 2.38 353.31 3 13.01 7.4 699 B2P34 0-10cm Fe,Fs,Pa 57.4
Block2 3.07 331.04 1 7.79 7.8 637 B2P35 0-10cm Fs,Qp 30.48
Block2 2.21 305.31 1 24.12 7.7 641 B2P36 0-10cm Fe,Pa 63.12
Block2 4.4 430.02 2 4.04 7.2 713 B2P37 0-10cm Ap,Fe,Fs,Qp 52.17
Block2 3.07 495.57 1 0.82 6.9 765 B2P38 0-10cm Ap,Fe,Fs 38.47
Block3 2.06 443 2 0 7 581 B3P1 0-10cm Qc,Qp 23.8
Block3 3.95 470 2 0 13.6 794 B3P2 0-10cm Oc,Qc,Qi,Qp 39.86
Block3 1.5 416 2 0 13.4 819 B3P3 0-10cm Oc,Qi 46.73
Block3 1.46 397 2 0 13.6 794 B3P4 0-10cm Qp 25.5
Block3 3.3 422 2 0 13.6 794 B3P5 0-10cm Oc,Qi,Qp 38.88
Block3 1 393 1 0 13.7 792 B3P6 0-10cm Qi 35.99
Block3 1 402 3 0 13.2 728 B3P7 0-10cm Cs 23.22
Block3 1.8 383 3 0 13.2 747 B3P8 0-10cm Oc,Qi 49.01
Block3 3 429 2 0 13.4 700 B3P9 0-10cm Oc,Qc,Qi 30.94
Block3 2.77 438 2 0 13 697 B3P10 0-10cm Cs,Qi,Qp 47.88
Block3 3.35 379 2 0 14 709 B3P11 0-10cm Cs,Oc,Qi,Qp 27.39
Block3 1.03 445 2 0 13 695 B3P12 0-10cm Cs 43.31
Block3 2.15 479 2 0 13 695 B3P13 0-10cm Cs,Qc 50.85
Block3 2.9 444 2 0 13.5 699 B3P14 0-10cm Cs,Qc,Qi 47.67
Block3 1 388 1 0 13.5 699 B3P15 0-10cm Qc 44.33
Block3 1.11 417 2 0 13.4 698 B3P16 0-10cm Qi 37.9
Block3 2.37 395 2 0 13.6 794 B3P17 0-10cm Qi,Qp 30.19
Block3 3.85 425 2 0 13.6 794 B3P18 0-10cm OC,Qc,Qi,Qp 40.14
Block3 2.02 478 2 0 13.3 793 B3P19 0-10cm Oc,Qc 41.85
Block3 2.55 508 2 0 13.3 792 B3P20 0-10cm Qc,Qi,Qp 30.98
Block3 1.94 464 2 0 13.4 700 B3P21 0-10cm Cs,Qi 77.66
Block3 3.66 410 3 0 13.7 707 B3P22 0-10cm Cs,Oc,Qc,Qi 41.56
Block3 3.43 523 2 0 12.8 691 B3P23 0-10cm Cs,Qc,Qi,Qp 39.08
Block3 1.86 416 2 0 13.2 694 B3P24 0-10cm Qc,Qi 43.99
Block3 2.45 355 2 0 13.4 700 B3P25 0-10cm Cs,OC,Qi 38.63
Block3 2.94 406 1 0 13.2 728 B3P26 0-10cm Cs,Qc,Qp 40.76
Block3 1.28 421 2 0 13.2 728 B3P27 0-10cm Qp 40.24
Block3 1.95 418 3 0 13.2 728 B3P28 0-10cm Cs,Qp 26.31
Block3 2.52 471 2 0 13.6 794 B3P29 0-10cm Oc,Qc,Qp 40.11
Block3 2.34 389 2 0 13.3 720 B3P30 0-10cm Cs,Oc 28.59
Block3 1.33 269 2 0 13.9 721 B3P31 0-10cm Oc 39.22
Block3 3.57 429 2 0 12.4 687 B3P32 0-10cm Cs,Oc,Qc,Qi, 34.04
Block3 3.74 519 2 0 12.4 687 B3P33 0-10cm Cs,Qc,Qi,Qp 42.68
Block3 4.74 480 3 0 12.4 687 B3P34 0-10cm Cs,Oc,Qc,Qp,Qi 40.75
Block3 1 254 2 0 14.1 731 B3P35 0-10cm Qc 36.14
Block3 2.47 436 2 0 13.2 728 B3P36 0-10cm Cs,Qc,Qp 28.27
Block4 1.12 182 1 97.71 7 581 B4P1 0-10cm Pa 41.27
Block4 1.5 157 1 4.37 6.9 585 B4P2 0-10cm Cb 24.79
Block4 1.13 163 1 97.37 6.9 580 B4P3 0-10cm Pa 29.14
Block4 3.66 171 1 26.56 6.8 582 B4P4 0-10cm Bp,Pa,Qr 27.01
Block4 2 176 1 0 6.9 576 B4P5 0-10cm Cb,Qr 23.39
Block4 3.1 190 1 42.44 6.8 585 B4P6 0-10cm Bp,Cb,Pa 29.98
Block4 3.8 190 1 35.89 6.8 585 B4P7 0-10cm Bp,Cb,Pa,Qr 32.7
Block4 2 180 1 0 6.8 585 B4P8 0-10cm Bp,Cb 23.3
Block4 2.18 195 1 64.67 6.8 584 B4P9 0-10cm Cb,Pa 27.27
Block4 1.74 145 1 2.92 6.8 582 B4P10 0-10cm Cb,Qr 28.13
Block4 1.75 185 1 0 6.8 581 B4P11 0-10cm Bp,Cb 24.78
Block4 1.23 160 1 5.27 6.8 583 B4P12 0-10cm Cb 23.16
Block4 2.94 160 1 0 6.9 578 B4P13 0-10cm Bp,Cb,Qr 26.84
Block4 3.65 150 1 70.67 6.9 578 B4P14 0-10cm Cb,Pa,Ps,Qr 40.99
Block4 2.95 184 1 33.63 6.8 583 B4P15 0-10cm Cb,Pa,Qr 40.3
Block4 2.35 186 1 45.55 6.8 583 B4P16 0-10cm Pa,Qr 52.01
Block4 3.95 155 1 57.64 6.9 586 B4P17 0-10cm Bp,Cb,Ps,Pa 59.3
Block4 3.23 160 1 46.08 6.9 581 B4P18 0-10cm Cb,Ps,Qr 21.05
Block4 2.16 175 1 97.97 6.9 582 B4P19 0-10cm Pa,Ps 26.99
Block4 1.73 173 1 1.72 7 576 B4P20 0-10cm Cb,Qr 36.26
Block4 1.41 170 1 91.64 6.9 577 B4P21 0-10cm Ps 36.91
Block4 3.34 160 1 71.93 6.9 581 B4P22 0-10cm Pa,Ps,Qr 30.89
Block4 2.89 170 1 78.58 6.9 582 B4P23 0-10cm Cb,Pa,Ps, 23.62
Block4 3.54 170 1 31.2 6.9 581 B4P24 0-10cm Cb,Ps,Qr 29.85
Block4 2.13 171 1 63.18 6.9 585 B4P25 0-10cm Cb,Ps 33.64
Block4 2.04 165 1 2.58 6.9 582 B4P26 0-10cm Bp,Cb 36.12
Block4 3.74 175 1 51.14 6.8 581 B4P27 0-10cm Cb,Pa,Ps,Qr 27.97
Block4 2.29 170 1 62.65 6.8 585 B4P28 0-10cm Bp,Pa 28.34
Block4 3.55 155 1 53.48 6.8 582 B4P29 0-10cm Bp,Ps,Qr 26.97
Block4 3.8 140 1 62.64 6.9 576 B4P30 0-10cm Bp,Cb,Ps,Pa 33.39
Block4 3.68 150 1 27.62 6.9 578 B4P31 0-10cm Bp,Cb,Pa,Qr 31.3
Block4 4.58 177 1 46.72 6.8 584 B4P32 0-10cm Bp,Pa,Ps,Qr 42.21
Block4 3.4 184 1 52.43 6.9 582 B4P33 0-10cm Bp,Cb,Ps 38.73
Block4 3.89 189 1 17.97 7 581 B4P34 0-10cm Bp,Cb,Ps,Qr 24.84
Block4 4.8 188 1 46.77 6.8 582 B4P35 0-10cm Bp,Cb,Pa,Ps,Qr 30.23
Block4 3.41 160 1 9.58 6.9 578 B4P36 0-10cm Bp,Cb,Qr 22.46
Block4 4.62 145 1 57.41 6.9 578 B4P37 0-10cm Bp,Cb,Pa,Ps,Qr 41.43
Block4 2.53 165 1 66.22 7 573 B4P38 0-10cm Bp,Cb,Ps 26.37
Block4 4.2 170 1 22.62 7 572 B4P39 0-10cm Bp,Cb,Ps,Qr 35.88
Block4 1.47 175 1 89.39 6.9 580 B4P40 0-10cm Ps 28.98
Block4 3.7 200 1 34.21 6.7 585 B4P41 0-10cm Bp,Pa,Ps,Qr 24.59
Block4 3.97 177 1 53.75 6.8 582 B4P42 0-10cm Bp,Cb,Pa,Ps, 29.49
Block4 3.28 186 1 74.3 7 570 B4P43 0-10cm Pa,Ps,Qr 23.52
Block5 1.17 838 2 98.88 5.8 675 B5P1 0-10cm Pa 44.37
Block5 1 865 2 100 5.8 675 B5P2 0-10cm Pa 44.63
Block5 3.14 869 2 25.95 5.8 675 B5P3 0-10cm Ap,Fs,Pa 58.27
Block5 2.08 1019 2 98.23 5.6 687 B5P5 0-10cm Aa,Pa 36.69
Block5 2.64 1045 2 8.05 5.6 687 B5P6 0-10cm Ap,Fs 36.08
Block5 1.12 1062 2 97.53 5.2 709 B5P7 0-10cm Aa 39.41
Block5 3.67 1028 2 40.83 5.4 701 B5P8 0-10cm Aa,Ap,Fs,Pa 35.27
Block5 2.37 984 2 45.68 5.4 701 B5P9 0-10cm Fs,Pa 45.6
Block5 3.13 968 2 78.33 5.4 701 B5P10 0-10cm Aa,Ap,Pa 42.73
Block5 3.08 805 2 69.69 5.8 678 B5P11 0-10cm Aa,Fs,Pa, 44.43
Block5 2 799 2 48.01 5.7 681 B5P12 0-10cm Fs,Pa 36.94
Block5 1 812 2 0 5.7 681 B5P13 0-10cm Fs 27.82
Block5 1.93 909 2 0 5.2 709 B5P14 0-10cm Ap,Fs 38.01
Block5 1.34 930 3 6.81 5.2 709 B5P15 0-10cm Ap 73.38
Block5 1.4 972 2 3.26 5.2 709 B5P16 0-10cm Ap 35.49
Block5 1.96 1047 2 97.16 4.6 742 B5P17 0-10cm Aa,Pa 43.19
Block5 2.97 1012 2 72.72 4.6 742 B5P18 0-10cm Aa,Fs,Pa, 39.22
Block5 2.18 951 2 30.92 5.7 681 B5P19 0-10cm Aa,Fs 31.47
Block5 2.94 869 2 26.36 5.7 681 B5P20 0-10cm Aa,Ap,Fs 31.62
Block5 2.94 718 2 18.25 5.7 681 B5P21 0-10cm Ap,Fs,Pa 38.54
Block5 2.89 843 2 75.15 6.2 655 B5P22 0-10cm Aa,Fs,Pa 34.27
Block5 1.06 894 2 98.99 5.6 688 B5P23 0-10cm Aa 33.8
Block5 1.96 919 2 59.5 5.6 688 B5P24 0-10cm Aa,Fs 32.04
Block5 3.12 1030 2 18.55 4.6 742 B5P25 0-10cm Aa,Ap,Fs 39.01
Block5 1.14 782 2 2.49 5.9 671 B5P26 0-10cm Fs 37.58
Block5 2.77 738 2 42.01 6.3 652 B5P27 0-10cm Ap,Pa 55.18
Block5 2.21 655 1 78.03 6.5 643 B5P28 0-10cm Aa,Fs 44
Block5 3.9 893 2 43.94 5.2 708 B5P29 0-10cm Aa,Ap,Fs,Pa 50.26
Block6 2.34 1224 2 61.84 9.7 545 B6P1 0-10cm Ps,Qf 74.58
Block6 3.09 1238 1 28.63 9.7 545 B6P2 0-10cm Pn,Ps,Qf 73.84
Block6 1.86 1228 2 31.43 9.7 549 B6P3 0-10cm Pn,Qf 86.92
Block6 1.85 1286 1 30.8 9.7 549 B6P4 0-10cm Ps,Qf 40.69
Block6 1.83 1283 1 70.89 9.7 549 B6P5 0-10cm Ps,Qf 67.12
Block6 1.93 1306 2 36.77 9.7 549 B6P6 0-10cm Pn,Qf 70.71
Block6 2.37 1291 2 10.51 9.7 548 B6P7 0-10cm Pn,Ps 64.21
Block6 2 1207 2 49.13 9.6 553 B6P8 0-10cm Pn,Qf 68.93
Block6 1 1211 2 100 9.6 553 B6P9 0-10cm Qf 38.62
Block6 1 1270 2 100 9.6 553 B6P10 0-10cm Qf 56.26
Block6 1 1187 2 100 9.9 537 B6P11 0-10cm Qf 47.39
Block6 1 1073 2 0 10 526 B6P12 0-10cm Pn 82.37
Block6 2.84 1010 2 73.39 10.8 491 B6P13 0-10cm Pn,Qf,Qi 98.04
Block6 2.17 999 2 33.5 10.8 491 B6P14 0-10cm Pn,Qi 84.62
Block6 1.1 980 2 1.87 10.8 491 B6P15 0-10cm Pn 33.39
Block6 2.12 1032 2 32.46 10.8 491 B6P16 0-10cm Pn,Qi 68.97
Block6 1.02 960 1 0.34 10.8 491 B6P17 0-10cm Pn 83.84
Block6 1 1403 1 0 9.1 573 B6P18 0-10cm Ps 46.3
Block6 1 1310 1 0 9.2 566 B6P19 0-10cm Ps 93.55
Block6 1 1311 1 0 9.1 569 B6P20 0-10cm Ps 95.7
Block6 1.98 1404 1 0 9 572 B6P21 0-10cm Pn,Ps 60.85
Block6 1.99 1325 1 0 9 570 B6P22 0-10cm Pn,Ps 75.42
Block6 2.32 1388 2 5.1 9.1 569 B6P23 0-10cm Pn,Ps 61.78
Block6 3.87 1377 1 45.55 9.1 557 B6P24 0-10cm Pn,Ps,Qf,Qi 82.94
Block6 3.58 1314 2 28.88 9.1 557 B6P25 0-10cm Pn,Ps,Qf 94.59
Block6 3.87 1387 2 47.63 9.1 557 B6P26 0-10cm Pn,Ps,Qf,Qi 92.18
Block6 2.98 1322 1 61.02 9.3 551 B6P27 0-10cm Pn,Qf,Qi 73.52
Block6 1.75 1360 1 100 9.4 541 B6P28 0-10cm Qf,Qi 47.47
Block6 3.44 1354 2 25.76 9.4 541 B6P29 0-10cm Pn,Ps,Qf,Qi 30.52
Block6 1.97 1350 1 100 9.4 541 B6P30 0-10cm Qf,Qi 37.82
Block6 1.85 1342 1 100 9.3 545 B6P31 0-10cm Qf,Qi 30.81
Block6 1 1236 2 100 10.3 504 B6P32 0-10cm Qi 64.2
Block6 1 1251 2 100 10.3 504 B6P33 0-10cm Qi 60.76
Block6 1.59 1250 2 100 10.7 484 B6P34 0-10cm Qf,Qi 30.09
Block6 2.57 1267 2 49.04 9.9 525 B6P35 0-10cm Pn,Qi 38.79
Block6 1.99 1211 2 44.14 10.2 511 B6P36 0-10cm Pn,Qi 60.9
答案 0 :(得分:4)
这适用于我sessionInfo()
,如下所示:
R Under development (unstable) (2014-09-17 r66626)
Platform: i686-pc-linux-gnu (32-bit)
other attached packages:
[1] glmulti_1.0.7 rJava_0.9-6 lme4_1.1-8 Rcpp_0.11.2 Matrix_1.1-4
loaded via a namespace (and not attached):
[1] compiler_3.2.0 grid_3.2.0 lattice_0.20-29 MASS_7.3-34
[5] minqa_1.2.3 nlme_3.1-117 nloptr_1.0.4 splines_3.2.0
[9] tools_3.2.0
使用以下代码:
library("lme4")
library("glmulti")
dd <- read.table("SO_glmulti.dat",header=TRUE)
m1 <- lmer(Yeild~ (TShann+Alt+Slope+CPT+MAT+MARF)^2+
(1|Blocks)+(1|Composition),
data=dd)
注意我在这里得到关于预测器缩放的警告 - 可能是无害的
setMethod('getfit', 'merMod', function(object, ...) {
summ <- coef(summary(object))
summ1 <- summ[,1:2,drop=FALSE]
## if (length(dimnames(summ)[[1]])==1) {
## summ1 <- matrix(summ1, nr=1,
## dimnames=list(c("(Intercept)"),
## c("Estimate","Std. Error")))
## }
cbind(summ1, df=rep(10000,length(fixef(object))))
})
这是glmulti
的旧版本 - 快速而肮脏,但取决于贬低公式。
lmer.glmulti<-function(formula,data,random="",...) {
lmer(paste(deparse(formula),random),data=data,
REML=FALSE,...)
}
更难以理解但更强大:
lmer.glmulti<-function(formula,data,random="",...) {
newf <- formula
newf[[3]] <- substitute(f+r,
list(f=newf[[3]],
r=reformulate(random)[[2]]))
lmer(newf,data=data,
REML=FALSE,...)
}
这就是我最终的结果:
glmulti_lmm <- glmulti(formula(m1,fixed.only=TRUE),
random="+(1|Blocks)+(1|Composition)",
data=dd,method="g",
deltaM=0.5,
fitfunc=lmer.glmulti,
intercept=TRUE,marginality=FALSE,level=2)
我最初尝试了默认method="h"
,在2650型号之后放弃了。我与method="g"
的第一次运行在50代之后得到了一个相当稳定的IC,但平均IC一直缓慢下降,所以我不耐烦并将deltaM
提升到0.5。
对于第一次运行,我得到IC = 1651.69761603866,模型为Yeild~1+CPT+CPT:TShann+CPT:Alt+MAT:Alt+MARF:CPT
在第二次运行时(deltaM增加),我有点幸运(IC = 1649.61044009369,Yeild~1+TShann+CPT+CPT:Alt+MAT:CPT+MARF:CPT
)。 (我不知道是否有办法设置种子/确保glmulti
的再现性)。该模型声称在120代之后趋同。
coef(glmulti_lmm)
对我来说很好。输出的底部(最高加权变量)是:
Estimate Uncond. variance Nb models Importance
[... skip ...]
CPT:MARF 1.334119e-03 5.491836e-07 11 0.7875438701
CPT:MAT 4.051261e-02 5.215084e-04 18 0.7995790960
TShann 1.260082e+00 1.650145e+00 35 0.8111493166
CPT -9.923303e-01 2.764638e-01 74 0.9600979205
Alt:CPT -2.465917e-04 7.937155e-09 72 0.9765910742
(Intercept) 3.754893e+01 5.988814e+01 100 1.0000000000
顺便说一下,你可能会对“生态学家过高估计预测变量在模型平均中的重要性感到兴趣:请求谨慎解释”,Galipaud等人。 http://dx.doi.org/10.1111/2041-210X.12251