Visualization tool for the Outlier Detection algorithm developed for the OpenBudegets project.

User Guide

Datasets List

In this module the user can see the list of datasets that have been already processed by the tool and which results are stored locally.

Clicking on an item of that list will render the details page for that dataset.

By activating the COMPARE MULTIPLE functionality, users can select various datasets and then click GO to open them all at the same time in a single chart for easy comparison. Selected datasets get a filled circled check mark as shown in the next image:

Each Dataset can be in two different statuses: Ready (green) and Processing (red), as shown in this picture:

Dataset Detail

Shows a bubble chart for the selected dataset. Each bubble is a transaction in the dataset as returned by the Outlier DM algorithm.

Dimensions

• X: Year.
• Y: Amount of money as it comes from the dataset itself.
• Size of the bubble: score obtained by this transaction telling us how "outlier" it is.
• Color of the bubble: budget phase or status in which this particular transaction appears in.

Hovering your mouse over a bubble will open a popup dialog with the exact information related to that transaction.

Adding a new Dataset

By clicking on the New Dataset button in the left navigation bar users are redirected to this view where they can tell the tool to import new datasets from OpenSpending, processing them through DAM using the outlier_dm algorithm and saving the results locally so other users or themselves can come later on an just visualize the results without computing everything again.

Field descriptions

• Dataset Name: name that will be shown in the datasets list page to refer to this dataset once it is added.
• Cube: dataset ID inside OpenSpending.
• FactsURI: Babbage API endpoint to be used.
• DamUrl: DAM endpoint used for executing the algorithm over the dataset.

Searching for a Dataset

• Clicking the SEARCH DATASET BY NAME button will open a search bar in which you can start typing part of the name of your dataset in the OpenSpending platform, the list will filter out all datasets not matching your search.
• After filtering, clicking on an element from that list will automatically paste the ID of that dataset into the "Cube" field and close the search component.
• After filling all the fields as you see fit, click the ADD DATASET button

Installation

Start inside Docker container

• Install docker and docker-compose in your machine.
• Run the backend app inside the webapp folder of this repository: (https://github.com/openbudgets/outlier_dm).
• Then: $ docker-compose up

If there's an error installing dependencies, make sure your /etc/resolv.conf file includes "nameserver 8.8.8.8" so the Docker daemon includes that DNS inside the container, then stop the container pressing Ctrl+C and executing docker-compose --build

Installing locally

# install backend dependencies and start backend server on a terminal (python 3)
cd backend
sudo pip3 install -f requirements.txt
python3 app.py

# install frontend dependencies and run web app
npm install
npm install -g cordova ionic
ionic serve --all

Navigate to http://localhost:8100/ in your browser.

Adding new languages for translation

• Open the src/app/pipes/translate.ts file and add the string translation for your new language in the shown format. For example, to add translations for French, one would extend every record in the translations dictionary. Each one of these records look like this:

'Dataset Details': {
  'en': 'Dataset Details',
  'de': 'Datensatzdetails',
  'fr': 'SomeTranslationInFrench
}

Where 'Dataset Details' is the original string and the others are translations to each respective language

• Then, open src/app/pages/configuration/configuration.ts and add your language label and value to the languages list, which is the one responsible for the SELECT element in the configuration page.