Closed Samuel-Campbell closed 6 years ago
Merging #433 into master will increase coverage by
0.14%
. The diff coverage is62.96%
.
@@ Coverage Diff @@
## master #433 +/- ##
=========================================
+ Coverage 70.96% 71.1% +0.14%
=========================================
Files 54 54
Lines 2273 2291 +18
Branches 24 24
=========================================
+ Hits 1613 1629 +16
- Misses 659 661 +2
Partials 1 1
Impacted Files | Coverage Δ | |
---|---|---|
src/ml_service/init.py | 83.33% <ø> (ø) |
:arrow_up: |
...n/single_output_regression/tenant_pays_landlord.py | 35.48% <0%> (ø) |
:arrow_up: |
...ture_extraction/post_processing/regex/regex_lib.py | 100% <100%> (ø) |
:arrow_up: |
...ion/single_output_regression/abstract_regressor.py | 50% <71.42%> (+5.4%) |
:arrow_up: |
...raining/classifier/multi_output/multi_class_svm.py | 86.84% <0%> (+1.97%) |
:arrow_up: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 6af610e...8172214. Read the comment docs.
1) Alternated activations functions between 'relu' and 'linear' 2) Removed outlier from dataset 3) Only train regressor based on important facts weighted by the classifier 4) Decreased epoch which was over-fitting data
R2: 0.72 absolute error: 302.01
R2 score is decresed from ~ 0.8 possible because less facts are given. --> I think more unimportant facts skew the results with false positivism
absolute error down from 377