Skip to content

Business intelligence made simple

Added experimental support for AnomalyDetection.rb

Andrew Kanec65f61c9d292
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 2355f226..4a7e7bf5 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,6 +1,7 @@
## 2.5.0 (unreleased)
- Added support for Slack OAuth tokens
+- Added experimental support for AnomalyDetection.rb
- Improved table preview for MySQL
- Fixed cohort analysis for MySQL
diff --git a/Gemfile b/Gemfile
index 15e69b13..a49a4671 100644
--- a/Gemfile
+++ b/Gemfile
@@ -15,5 +15,6 @@ gem "mysql2"
gem "sqlite3"
# anomaly detection and forecasting
+gem "anomaly_detection" if ENV["TEST_ANOMALY_DETECTION"]
gem "prophet-rb" if ENV["TEST_PROPHET"]
gem "trend" if ENV["TEST_TREND"]
diff --git a/README.md b/README.md
index e71531fa..8f5d963b 100644
--- a/README.md
+++ b/README.md
@@ -456,46 +456,20 @@ For the [self-hosted API](https://github.com/ankane/trend-api), create an initia
Trend.url = "http://localhost:8000"
```
-### R
+### AnomalyDetection.rb (unreleased, experimental)
-R uses Twitter’s [AnomalyDetection](https://github.com/twitter/AnomalyDetection) library.
+Add [anomaly_detection](https://github.com/ankane/AnomalyDetection.rb) to your Gemfile:
-First, [install R](https://cloud.r-project.org/). Then, run:
-
-```R
-install.packages("remotes")
-remotes::install_github("twitter/AnomalyDetection")
+```ruby
+gem 'anomaly_detection'
```
And add to `config/blazer.yml`:
```yml
-anomaly_checks: r
-```
-
-If upgrading from version 1.4 or below, also follow the [upgrade instructions](#15).
-
-If you’re on Heroku, follow the additional instructions below.
-
-### R on Heroku
-
-Add the [R buildpack](https://github.com/virtualstaticvoid/heroku-buildpack-r) to your app.
-
-```sh
-heroku buildpacks:add --index 1 https://github.com/virtualstaticvoid/heroku-buildpack-r.git
+anomaly_checks: anomaly_detection
```
-And create an `init.R` with:
-
-```r
-if (!"AnomalyDetection" %in% installed.packages()) {
- install.packages("remotes")
- remotes::install_github("twitter/AnomalyDetection")
-}
-```
-
-Commit and deploy away. The first deploy may take a few minutes.
-
## Forecasting
Blazer supports for two different forecasting methods. [Example](https://blazer.dokkuapp.com/queries/18-forecast?forecast=t)
diff --git a/lib/blazer/result.rb b/lib/blazer/result.rb
index 27490d41..8456ea7c 100644
--- a/lib/blazer/result.rb
+++ b/lib/blazer/result.rb
@@ -189,6 +189,10 @@ module Blazer
when "trend"
anomalies = Trend.anomalies(Hash[series])
anomalies.include?(series.last[0])
+ when "anomaly_detection"
+ period = 7 # TODO determine period
+ anomalies = AnomalyDetection.detect(Hash[series], period: period)
+ anomalies.include?(series.last[0])
else
csv_str =
CSV.generate do |csv|
diff --git a/lib/generators/blazer/templates/config.yml.tt b/lib/generators/blazer/templates/config.yml.tt
index eddde94e..66abf6cc 100644
--- a/lib/generators/blazer/templates/config.yml.tt
+++ b/lib/generators/blazer/templates/config.yml.tt
@@ -63,7 +63,7 @@ check_schedules:
# enable anomaly detection
# note: with trend, time series are sent to https://trendapi.org
-# anomaly_checks: prophet / trend / r
+# anomaly_checks: prophet / trend / anomaly_detection
# enable forecasting
# note: with trend, time series are sent to https://trendapi.org
diff --git a/test/checks_test.rb b/test/checks_test.rb
index c1f1ab7b..6832a924 100644
--- a/test/checks_test.rb
+++ b/test/checks_test.rb
@@ -61,6 +61,12 @@ class ChecksTest < ActionDispatch::IntegrationTest
assert_anomaly("trend")
end
+ def test_anomaly_anomaly_detection
+ skip unless ENV["TEST_ANOMALY_DETECTION"]
+
+ assert_anomaly("anomaly_detection")
+ end
+
def test_emails
query = create_query
check = create_check(query: query, check_type: "bad_data", emails: "[email protected],[email protected]")