silvermine / dynamodb-capacity-manager
70%
master: 70%

LAST BUILD BRANCH: MrMarCode/upgrade-to-node-20
DEFAULT BRANCH: master
Repo Added
Files 22
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

LAST BUILD ON BRANCH fix_nan_forecast
branch: fix_nan_forecast
CHANGE BRANCH
x

107

Pull #17

travis-ci

web-flow 
Use seconds rather than milliseconds for x-axis of regression

When we were using milliseconds as the x-axis for the regression, there were
some rare cases in which the regression function would return NaN. The sample
data added as a test in this commit is one of these cases. With the sample
data, this was occuring because the `rhs` input to the [`gaussianElimination`
function][1] was:

```js
[
   [ 20, 30424398600000, 4.628220150858649e+25 ],
   [ 30424398600000, 4.628220150858649e+25, 7.040540733914512e+37 ],
   [ 4.628220150858649e+25, 7.040540733914512e+37, 1.0710210882409248e+50 ],
   [ 140.33333333333337, 213477862297500, 3.2474677689629145e+26 ],
]
```

This then resulted in the following chain of calculations occuring [here][2]:

```
-7.310223315482211e+24 = 7.040540733914512e+37 - (1.0710210882409248e+50 * 30424398600000 / 4.628220150858649e+25)
-2396591751168 = 4.628220150858649e+25 - (7.040540733914512e+37 * 30424398600000 / 4.628220150858649e+25)
-9603546873856 = 4.628220150858649e+25 - (1.0710210882409248e+50 * 20 / 4.628220150858649e+25)
-3.1484375 = 30424398600000 - (7.040540733914512e+37 * 20 / 4.628220150858649e+25)
0 = -9603546873856 - (-7.310223315482211e+24 * -3.1484375 / -2396591751168)
```

The resulting 0 then causes a divide by zero error [here][3]. WolframAlpha
reports that the value for the last calculation should have actually been
`0.0002748272381722927093505859375`.

```
x = -9603546873856 - (-7.310223315482211e+24 * -3.1484375 / -2396591751168)
x = -9603546873856 - (-7310223315482211000000000 * -3.1484375 / -2396591751168)
x = -9603546873856 - (23015781219838523695312500 / -2396591751168)
x = -9603546873856 - (-9603546873856.0002748272381722927093505859375)
x = -9603546873856 + 9603546873856.0002748272381722927093505859375
x = 0.0002748272381722927093505859375
```

By using a scale of seconds for the x-axis regression, the following is used
for the `rhs` for the `gaussianElimination` function:

```
[
   [ 20, 30424398600,... (continued)
Pull Request #17: Prevent sudden drop in capacity when forecast is NaN

11 of 11 new or added lines in 2 files covered. (100.0%)

221 of 317 relevant lines covered (69.72%)

34.12 hits per line

Relevant lines Covered
317 RELEVANT LINES 221 COVERED LINES
34.12 HITS PER LINE
Source Files on fix_nan_forecast
Detailed source file information is not available for this build.

Recent builds

Builds Branch Commit Type Ran Committer Via Coverage
107 fix_nan_forecast Use seconds rather than milliseconds for x-axis of regression When we were using milliseconds as the x-axis for the regression, there were some rare cases in which the regression function would return NaN. The sample data added as a test in this ... Pull #17 web-flow travis-ci pending completion  
106 fix_nan_forecast Use seconds rather than milliseconds for x-axis of regression When we were using milliseconds as the x-axis for the regression, there were some rare cases in which the regression function would return NaN. The sample data added as a test in this ... push onebytegone travis-ci pending completion  
104 fix_nan_forecast Use seconds rather than milliseconds for x-axis of regression When we were using milliseconds as the x-axis for the regression, there were some rare cases in which the regression function would return NaN. The sample data added as a test in this ... push onebytegone travis-ci pending completion  
98 fix_nan_forecast Use seconds rather than milliseconds for x-axis of regression When we were using milliseconds as the x-axis for the regression, there were some rare cases in which the regression function would return NaN. The sample data added as a test in this ... push onebytegone travis-ci pending completion  
94 fix_nan_forecast Use seconds rather than milliseconds for x-axis of regression When we were using milliseconds as the x-axis for the regression, there were some rare cases in which the regression function would return NaN. The sample data added as a test in this ... push onebytegone travis-ci pending completion  
81 fix_nan_forecast Use seconds rather than milliseconds for x-axis of regression When we were using milliseconds as the x-axis for the regression, there were some rare cases in which the regression function would return NaN. The sample data added as a test in this ... push onebytegone travis-ci pending completion  
80 fix_nan_forecast Use seconds rather than milliseconds for x-axis of regression When we were using milliseconds as the x-axis for the regression, there were some rare cases in which the regression function would return NaN. The sample data added as a test in this ... push onebytegone travis-ci pending completion  
79 fix_nan_forecast Use seconds rather than milliseconds for x-axis of regression When we were using milliseconds as the x-axis for the regression, there were some rare cases in which the regression function would return NaN. The sample data added as a test in this ... push onebytegone travis-ci pending completion  
See All Builds (132)