emergent-analytics / workstreams Goto Github PK

Provide docker setup for Economic Engine

Provisioning a docker image helps a lot in creating a reproducible, well configured environment.

Workstream

WS3, in particular Simulation engine app

Your environment

• ubuntu 20/AMD/Intel and macOS Catalina 10.15
• docker snap-in, docker app

Steps to reproduce

N/A

Expected behaviour

• Allow docker container build straight from the repo
• Amend README.md

Actual behaviour

N/A

Subject of the issue

There are multiple wave zones overlapping each other displayed in the case data bar chart.

Workstream

Workstream 1/cookiecutter labelling tool

Your environment

• docker

Steps to reproduce

Open up cookiecutter, you will see multiple red-green wave/calm zones for one country, at country level

Expected behaviour

Changepoint detection should not result in overlapping zones of wave/calm

Actual behaviour

The query to retrieve computes zones does not consider the data source selected, and will, at country level, retrieve zones from both Johns Hopkins global, and ECDC datasets. Adding the currently selected data source to the WHERE clause in the could should revert back to desired behaviour, only displaying computed waves pertinent to the selected dataset, with no overlaps.

Subject of the issue

Some case number datasets displayed by cookiecutter may compute out of range datetimes well before 2020-01-01.

Workstream

Workstream 1

Your environment

• using the docker versions of python and libraries

Steps to reproduce

This may vary day by day as it is caused by numerical instabilities of some of the wave detectors

Expected behaviour

x axis/datetime should only display relevant time ranges

Actual behaviour

Display zooms out and displays data from 1980-01-01/1970-01-01

Subject of the issue

https://www.ecdc.europa.eu/en/publications-data/download-todays-data-geographic-distribution-covid-19-cases-worldwide

From the website

ECDC switched to a weekly reporting schedule for the COVID-19 situation worldwide and in the EU/EEA and the UK on 17 December this year. Hence, all daily updates have been discontinued from 14 December. ECDC will publish updates on the number of cases and deaths reported worldwide and aggregated by week every Thursday. The weekly data will be available as downloadable files in the following formats: XLSX, CSV, JSON and XML. As an exception, the weekly updates for the end-of-year festive season will be published on 23 December and 30 December 2020.

Workstream

Workstream 1, cookicutter labelling

Your environment

• N/A, backend

Steps to reproduce

jupyter notebook Download Case Data will fail as also the column names have changed.

Expected behaviour

European authorities to continue reporting daily numbers, maybe at a reduced frequency.

Actual behaviour

ECDC decided something else.

Subject of the issue

Docker compose build default variable definitions were missing

Workstream

ws1/ labelling tool/ cookiecutter

Your environment

• N/A, generic

Steps to reproduce

When building/running the container services, a missing POSTGRES_PORT definition is being flagged and the container cannot be run

Expected behaviour

Give an example .env file and instruct how to use it

Actual behaviour

No .env file available as a template

Subject of the issue

The cookiecutter tool suite is very complex and some features are not obvious at first glance.

Your environment

• N/A

Steps to reproduce

Directly connect to the bokeh application.

Expected behaviour

The features of the tool and their purposes ought to be described.

Actual behaviour

Users are not able to expliot the toolset.

Subject of the issue

A lot of rework was done to move the data storage concept from files to SQL backend. This branch has now matured and can be pulled back into master.

Workstream

Cookiecutter supports ws1 (health) and ws3 (economic engine)

Your environment

• python 3.6 and 3.8
• ubuntu 16 and 18
• Chrome Canary, Chrome

Steps to reproduce

N/A

Expected behaviour

Should work as specified

Actual behaviour

N/A

Subject of the issue

The heatmap with the temporal clusters is not updated for certain countries.

Workstream

ws1, cookiecutter labelling tool

Your environment

• N/A

Steps to reproduce

On cookiecutter,

• look at the landing page (which shows Germany by default), it will display a correct heatmap for the Temporal clusters
• select another country for which stringency cluster data would be available. e.g. United Kingdom,
• the heatmap will shrink and no data will be displayed
• changing back to Germany, the display will function

Expected behaviour

Stringency cluster data should be displayed when available

Actual behaviour

The code was reassigning an newly created range to the y axis (the range is the country names, which needs updating. This is not updating the parent figure.

Instead of

self.p_oxcluster.y_range = FactorRange(factors=sorted(df.country.unique(),reverse=True))

the correct assignment is to the factors attribute

self.p_oxcluster.y_range.factors=sorted(df.country.unique(),reverse=True)