cv error plot for lasso

cv error plot
cv error plot

lambda_lse, lambda_min

hetamap
hetamap

Sample Size Experiment

Basic Experimental Setup

  • n: 100 to 1000, incrementing by 100
  • p: Fixed at 1000
  • Number of signal variables: 3
  • Signal strength: 1
  • Intercept: 1
  • Error term: N(0,1)
  • Design matrix Xij : N(0,1)

only global seed : set.seed(42)

Overall RMSE
Overall RMSE

Signal RMSE null RMSE selected value

## Rows: 10 Columns: 5
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## dbl (5): n, Lasso_min, Debiased_min, Lasso_1se, Debiased_1se
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 10 Columns: 5
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## dbl (5): n, Lasso_min, Debiased_min, Lasso_1se, Debiased_1se
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 10 Columns: 5
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## dbl (5): n, Lasso_min, Debiased_min, Lasso_1se, Debiased_1se
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## Rows: 10 Columns: 5
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## dbl (5): n, Lasso_min, Debiased_min, Lasso_1se, Debiased_1se
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## # A tibble: 10 × 5
##        n Lasso_min Debiased_min Lasso_1se Debiased_1se
##    <dbl>     <dbl>        <dbl>     <dbl>        <dbl>
##  1   100    0.0199       0.0999    0.0291       0.121 
##  2   200    0.0109       0.0723    0.0161       0.0808
##  3   300    0.0111       0.0592    0.0139       0.0635
##  4   400    0.0077       0.0511    0.011        0.0528
##  5   500    0.0084       0.0472    0.0115       0.0497
##  6   600    0.0057       0.0415    0.0083       0.0426
##  7   700    0.0047       0.0389    0.0098       0.0406
##  8   800    0.0064       0.0338    0.0085       0.0344
##  9   900    0.0053       0.0334    0.0084       0.0342
## 10  1000    0.0062       0.0299    0.0069       0.0318
## # A tibble: 10 × 5
##        n Lasso_min Debiased_min Lasso_1se Debiased_1se
##    <dbl>     <dbl>        <dbl>     <dbl>        <dbl>
##  1   100    0.357        0.152      0.531       0.16  
##  2   200    0.193        0.0756     0.293       0.0692
##  3   300    0.195        0.0954     0.253       0.0871
##  4   400    0.139        0.0245     0.201       0.0268
##  5   500    0.148        0.0387     0.210       0.0393
##  6   600    0.104        0.0324     0.151       0.0337
##  7   700    0.0825       0.0304     0.180       0.0278
##  8   800    0.114        0.0261     0.154       0.0247
##  9   900    0.095        0.021      0.153       0.0224
## 10  1000    0.0825       0.0359     0.125       0.0338
## # A tibble: 10 × 5
##        n Lasso_min Debiased_min Lasso_1se Debiased_1se
##    <dbl>     <dbl>        <dbl>     <dbl>        <dbl>
##  1   100    0.0036       0.0998    0            0.121 
##  2   200    0.0027       0.0723    0            0.0809
##  3   300    0.003        0.0591    0.0001       0.0634
##  4   400    0.0009       0.0512    0            0.0528
##  5   500    0.0023       0.0473    0            0.0497
##  6   600    0.0008       0.0415    0            0.0426
##  7   700    0.0013       0.0389    0            0.0407
##  8   800    0.0013       0.0338    0            0.0345
##  9   900    0.0009       0.0334    0            0.0342
## 10  1000    0.0043       0.0299    0.0008       0.0318
## # A tibble: 10 × 5
##        n Lasso_min Debiased_min Lasso_1se Debiased_1se
##    <dbl>     <dbl>        <dbl>     <dbl>        <dbl>
##  1   100    0.0036       0.0998    0            0.121 
##  2   200    0.0027       0.0723    0            0.0809
##  3   300    0.003        0.0591    0.0001       0.0634
##  4   400    0.0009       0.0512    0            0.0528
##  5   500    0.0023       0.0473    0            0.0497
##  6   600    0.0008       0.0415    0            0.0426
##  7   700    0.0013       0.0389    0            0.0407
##  8   800    0.0013       0.0338    0            0.0345
##  9   900    0.0009       0.0334    0            0.0342
## 10  1000    0.0043       0.0299    0.0008       0.0318

independent seed : set.seed(42)

overall RMSE signal RMSE null RMSE

selected value
selected value
## Rows: 4 Columns: 13
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr  (1): Method
## dbl (12): Overall_Mean, Overall_SD, Overall_Min, Overall_Max, Signal_Mean, S...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## # A tibble: 4 × 13
##   Method   Overall_Mean Overall_SD Overall_Min Overall_Max Signal_Mean Signal_SD
##   <chr>           <dbl>      <dbl>       <dbl>       <dbl>       <dbl>     <dbl>
## 1 Debiase…      0.0547     0.0263      0.0327       0.113       0.0432    0.0285
## 2 Debiase…      0.0478     0.0164      0.0316       0.0790      0.0412    0.0282
## 3 Lasso_1…      0.0110     0.00338     0.00633      0.0165      0.199     0.0600
## 4 Lasso_m…      0.00755    0.00344     0.00438      0.0155      0.120     0.0408
## # ℹ 6 more variables: Signal_Min <dbl>, Signal_Max <dbl>, Null_Mean <dbl>,
## #   Null_SD <dbl>, Null_Min <dbl>, Null_Max <dbl>

monte_carlo : avg of 50 seeds

overall RMSE
overall RMSE
signal RMSE
signal RMSE
null RMSE
null RMSE
selected value
selected value
## Rows: 4 Columns: 13
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr  (1): Method
## dbl (12): Overall_Mean, Overall_SD, Overall_Min, Overall_Max, Signal_Mean, S...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## # A tibble: 4 × 13
##   Method   Overall_Mean Overall_SD Overall_Min Overall_Max Signal_Mean Signal_SD
##   <chr>           <dbl>      <dbl>       <dbl>       <dbl>       <dbl>     <dbl>
## 1 Debiase…       0.0547     0.0263      0.0327      0.113       0.0432    0.0285
## 2 Debiase…       0.0478     0.0164      0.0316      0.079       0.0412    0.0282
## 3 Lasso_1…       0.011      0.0034      0.0063      0.0165      0.199     0.06  
## 4 Lasso_m…       0.0075     0.0034      0.0044      0.0155      0.120     0.0408
## # ℹ 6 more variables: Signal_Min <dbl>, Signal_Max <dbl>, Null_Mean <dbl>,
## #   Null_SD <dbl>, Null_Min <dbl>, Null_Max <dbl>

Sparsity Experiment

Basic Experimental Setup

  • n: Fixed at 1000
  • p: Fixed at 1000
  • Number of signal variables: (3, 5, 10, 15, 20, 30, 40, 50)
  • Signal strength: 1
  • Intercept: 1
  • Error term: N(0,1)
  • Design matrix Xij : N(0,1)

only global seed : set.seed(42)

overall RMSE signal RMSE

null RMSE
null RMSE
selected value
selected value
## Rows: 4 Columns: 13
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr  (1): Method
## dbl (12): Overall_Mean, Overall_SD, Overall_Min, Overall_Max, Signal_Mean, S...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## # A tibble: 4 × 13
##   Method   Overall_Mean Overall_SD Overall_Min Overall_Max Signal_Mean Signal_SD
##   <chr>           <dbl>      <dbl>       <dbl>       <dbl>       <dbl>     <dbl>
## 1 Debiase…       0.0333     0.0009      0.0321      0.0348      0.0286    0.0099
## 2 Debiase…       0.0312     0.0012      0.03        0.0333      0.0291    0.0097
## 3 Lasso_1…       0.0132     0.0041      0.007       0.0196      0.103     0.0205
## 4 Lasso_m…       0.0107     0.0047      0.0041      0.0183      0.0728    0.0098
## # ℹ 6 more variables: Signal_Min <dbl>, Signal_Max <dbl>, Null_Mean <dbl>,
## #   Null_SD <dbl>, Null_Min <dbl>, Null_Max <dbl>

monte_carlo : avg of 50 seeds

overall RMSE
overall RMSE
signal RMSE
signal RMSE
null RMSE
null RMSE
selected value
selected value

Signal Strength experiment

Basic Experimental Setup

  • n: Fixed at 1000
  • p: Fixed at 1000
  • Number of signal variables: 3
  • Signal strength: (0.3, 0.5, 1, 2, 3, 4, 5)
  • Intercept: 1
  • Error term: N(0,1)
  • Design matrix Xij : N(0,1)

only global seed : set.seed(42)

overall RMSE
overall RMSE
signal RMSE
signal RMSE
null RMSE
null RMSE
selected value
selected value
## Rows: 4 Columns: 13
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr  (1): Method
## dbl (12): Overall_Mean, Overall_SD, Overall_Min, Overall_Max, Signal_Mean, S...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## # A tibble: 4 × 13
##   Method   Overall_Mean Overall_SD Overall_Min Overall_Max Signal_Mean Signal_SD
##   <chr>           <dbl>      <dbl>       <dbl>       <dbl>       <dbl>     <dbl>
## 1 Debiase…      0.0327     0.00160     0.0308      0.0350       0.0298    0.0115
## 2 Debiase…      0.0315     0.00165     0.0295      0.0342       0.0289    0.0114
## 3 Lasso_1…      0.00747    0.00152     0.00539     0.00978      0.135     0.0292
## 4 Lasso_m…      0.00508    0.00110     0.00352     0.00641      0.0778    0.0169
## # ℹ 6 more variables: Signal_Min <dbl>, Signal_Max <dbl>, Null_Mean <dbl>,
## #   Null_SD <dbl>, Null_Min <dbl>, Null_Max <dbl>

monte_carlo : avg of 50 seeds

overall RMSE
overall RMSE
signal RMSE
signal RMSE
null RMSE
null RMSE
selected value
selected value
## Rows: 7 Columns: 5
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## dbl (5): SignalStrength, Debiased_1se, Debiased_min, Lasso_1se, Lasso_min
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## # A tibble: 7 × 5
##   SignalStrength Debiased_1se Debiased_min Lasso_1se Lasso_min
##            <dbl>        <dbl>        <dbl>     <dbl>     <dbl>
## 1            0.3         3.22         18.5      3.22      18.5
## 2            0.5         3.12         16.3      3.12      16.3
## 3            1           3            14.3      3         14.3
## 4            2           3.24         21.2      3.24      21.2
## 5            3           3.06         15.8      3.06      15.8
## 6            4           3.18         19.1      3.18      19.1
## 7            5           3.24         16.2      3.24      16.2

Compound symmetry experiment

Basic Experimental Setup

  • n: Fixed at 1000
  • p: Fixed at 1000
  • Number of signal variables: 3
  • Signal strength: 1
  • Intercept: 1
  • Error term: N(0,1)
  • Design matrix Xij : Compound symmetry

only global seed : set.seed(42)

overall RMSE signal RMSE

null RMSE
null RMSE
## Rows: 36 Columns: 5
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): Method
## dbl (4): Overall_RMSE, Signal_RMSE, Null_RMSE, rho
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## # A tibble: 36 × 5
##    Method       Overall_RMSE Signal_RMSE Null_RMSE   rho
##    <chr>               <dbl>       <dbl>     <dbl> <dbl>
##  1 Lasso_min         0.00508      0.0827   0.00231   0.1
##  2 Debiased_min      0.0311       0.0302   0.0311    0.1
##  3 Lasso_1se         0.00905      0.165    0         0.1
##  4 Debiased_1se      0.0377       0.0348   0.0377    0.1
##  5 Lasso_min         0.00335      0.0567   0.00126   0.2
##  6 Debiased_min      0.0324       0.0515   0.0323    0.2
##  7 Lasso_1se         0.00766      0.140    0         0.2
##  8 Debiased_1se      0.0478       0.0726   0.0477    0.2
##  9 Lasso_min         0.00499      0.0822   0.00216   0.3
## 10 Debiased_min      0.0338       0.0280   0.0338    0.3
## # ℹ 26 more rows

monte_carlo : avg of 50 seeds

overall RMSE signal RMSE

null RMSE
null RMSE
## Rows: 36 Columns: 8
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): Method
## dbl (7): rho, Overall_RMSE.mean, Overall_RMSE.sd, Signal_RMSE.mean, Signal_R...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
## # A tibble: 36 × 8
##    Method         rho Overall_RMSE.mean Overall_RMSE.sd Signal_RMSE.mean
##    <chr>        <dbl>             <dbl>           <dbl>            <dbl>
##  1 Debiased_1se   0.1           0.0421          0.00465           0.0330
##  2 Debiased_min   0.1           0.0322          0.00149           0.0294
##  3 Lasso_1se      0.1           0.00728         0.00117           0.133 
##  4 Lasso_min      0.1           0.00483         0.00102           0.0799
##  5 Debiased_1se   0.2           0.0464          0.00461           0.0362
##  6 Debiased_min   0.2           0.0331          0.00124           0.0311
##  7 Lasso_1se      0.2           0.00683         0.00107           0.124 
##  8 Lasso_min      0.2           0.00502         0.00108           0.0820
##  9 Debiased_1se   0.3           0.0498          0.00502           0.0391
## 10 Debiased_min   0.3           0.0343          0.00119           0.0335
## # ℹ 26 more rows
## # ℹ 3 more variables: Signal_RMSE.sd <dbl>, Null_RMSE.mean <dbl>,
## #   Null_RMSE.sd <dbl>

Sample Size Experiment