|
Size: 810
Comment:
|
Size: 2509
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 2: | Line 2: |
| = Results for Exercise Sheet 12 (statistical significance tests) = Please add your row to the table below, following the examples already there. Report test results for each of the three test datasets in increasing size (50 examples/200 examples/all examples). Specify all numbers in percent, rounded to the first decimal place. Briefly specify your variant and any refinements you might have implemented. |
|
| Line 3: | Line 5: |
| '''Precision''': the precision in percent rounded to integer | ||<tablewidth="100%" tableheight="auto" style="text-align:center"> '''Name''' ||<:> '''Baseline Precisions''' ||<:> '''Variant Precisions''' ||<:>'''p-values t-Test''' ||'''Additional Information''' || ||Elmar ||<:>74.0% / 66.0% / 69.8%||<:>80.0% / 68.0% / 73.2%||<:>47.8% / 67.1% / 0.2% || Variant: logistic regression without batches, 10 iterations || ||Elias & Maxi ||<:>86.4% / 63.8% / 71.4%||<:>84.6% / 63.9% / 72.1%||<:>82.7% / 83.7% / 26.9% || Variant: logistic regression (no batches), 30 iterations || ||ES ||<:>68.0% / 64.0% / 71.1%||<:>74.0% / 69.0% / 73.1%||<:>64.0% / 45.3% / 21.0% || Variant: logistic regression (no batches), stop iteration if wrong predictions are less then 7% of documents, averaging || ||ID ||<:>68.0% / 65.0% / 68.2%||<:>76.0% / 66.0% / 72.4%||<:>52,8% / 88.2% / 1.04% || Variant: logistic regression without batches, 10 iterations + averaging || ||Hui Hui ||<:>74.0% / 66.0% / 69.8%||<:>78.0% / 67.0% / 71.3%||<:>69.5% / 81.8% / 15.9% || Variant: logistic regression || ||Perspective Daily ||<:>74.0% / 66.0% / 69.8%||<:>76.0% / 68.0% / 72.8%||<:>43.5% / 38.1% / 2.7% || Variant: Averaging || ||Alex ||<:>74.0% / 67.0% / 69.0%||<:>76.0% / 70.0% / 74.0%||<:>87.1% / 70.4% / 0.71% || Variant: logistic regression with averaging, 10 Iterations || ||David ||<:>75.2% / 67.0% / 69.7%||<:>76.9% / 67.3% / 72.8%||<:>47.3% / 49.2% / 16.9% || Variant: Averaging, 10 Iterations || |
| Line 5: | Line 15: |
| '''p-value (Z)''': the p-value of the Z-tests in percent rounded to integer '''p-value (T)''': the p-value of the T-tests in percent rounded to integer ||<tablewidth="100%" tableheight="auto" style="text-align:center" |2> '''Name''' ||<:> '''Baseline''' |||||| '''Variant 1''' |||||| '''Variant 2''' ||<style="text-align:center" |2>'''Parameters''' || ||<:> '''Precision''' ||<:> '''Precision''' ||<:> '''p-value (Z)''' ||<:> '''p-value (T)''' ||<:> '''Precision''' ||<:> '''p-value (Z)''' ||<:> '''p-value (T)''' || ||Elmar ||<:>50 / 60 / 70||<:>50 / 60 / 72||<:>1.5 / 4.6 / 2 ||<:>1.5 / 4.6 / 2 ||<:>50 / 60 / 73||<:>1.5 / 4.6 / 0 ||<:>1.5 / 4.6 / 0 || V1: avg. perceptron, V2: log. regression || |
||<tablewidth="100%" tableheight="auto" style="text-align:center" |2> '''Name''' |||||| '''50 Examples''' |||||| '''200 Examples''' |||||| '''3140 Examples''' ||<style="text-align:center" |2>'''Parameters''' || ||<:> '''P@3''' ||<style="text-align:center"> '''P@R''' ||<:> '''MAP''' ||<:> '''P@3''' ||<:> '''P@R''' ||<:> '''MAP''' ||<:> '''P@3''' ||<:> '''P@R''' ||<:> '''MAP''' || ||Elmar ||<:> 50.0 ||<:> 39.0 ||<:> 37.5 ||<:> 3.3 ||<:> 1.9 ||<:> 1.4 ||<:> 56.7 ||<:> 37.9 ||<:> 37.8 || bm25k=1.75, bm25b=0.75, lambda=0.67, k=10, m=20000 || |
Results for Exercise Sheet 12 (statistical significance tests)
Please add your row to the table below, following the examples already there. Report test results for each of the three test datasets in increasing size (50 examples/200 examples/all examples). Specify all numbers in percent, rounded to the first decimal place. Briefly specify your variant and any refinements you might have implemented.
Name |
Baseline Precisions |
Variant Precisions |
p-values t-Test |
Additional Information |
Elmar |
74.0% / 66.0% / 69.8% |
80.0% / 68.0% / 73.2% |
47.8% / 67.1% / 0.2% |
Variant: logistic regression without batches, 10 iterations |
Elias & Maxi |
86.4% / 63.8% / 71.4% |
84.6% / 63.9% / 72.1% |
82.7% / 83.7% / 26.9% |
Variant: logistic regression (no batches), 30 iterations |
ES |
68.0% / 64.0% / 71.1% |
74.0% / 69.0% / 73.1% |
64.0% / 45.3% / 21.0% |
Variant: logistic regression (no batches), stop iteration if wrong predictions are less then 7% of documents, averaging |
ID |
68.0% / 65.0% / 68.2% |
76.0% / 66.0% / 72.4% |
52,8% / 88.2% / 1.04% |
Variant: logistic regression without batches, 10 iterations + averaging |
Hui Hui |
74.0% / 66.0% / 69.8% |
78.0% / 67.0% / 71.3% |
69.5% / 81.8% / 15.9% |
Variant: logistic regression |
Perspective Daily |
74.0% / 66.0% / 69.8% |
76.0% / 68.0% / 72.8% |
43.5% / 38.1% / 2.7% |
Variant: Averaging |
Alex |
74.0% / 67.0% / 69.0% |
76.0% / 70.0% / 74.0% |
87.1% / 70.4% / 0.71% |
Variant: logistic regression with averaging, 10 Iterations |
David |
75.2% / 67.0% / 69.7% |
76.9% / 67.3% / 72.8% |
47.3% / 49.2% / 16.9% |
Variant: Averaging, 10 Iterations |
Name |
50 Examples |
200 Examples |
3140 Examples |
Parameters |
||||||
P@3 |
P@R |
MAP |
P@3 |
P@R |
MAP |
P@3 |
P@R |
MAP |
||
Elmar |
50.0 |
39.0 |
37.5 |
3.3 |
1.9 |
1.4 |
56.7 |
37.9 |
37.8 |
bm25k=1.75, bm25b=0.75, lambda=0.67, k=10, m=20000 |