Troubleshoot Dataset Download

Broken download link

If the button “Download scenario X” does not redirected to a Dropbox download page, then the link is broken. Please report the incident to a team member and it should be solved in less than a day. 

Download file is corrupted

Scenario zip files can get corrupted during download. The longer the download goes on, the higher the chances of it getting corrupted. Two alternatives to reduce the chances of the download getting corrupted are presented below.
Decrease the download size by downloading files in parts
If the scenario you are attempting to download is scenario 36 or newer, make sure to download the file parts individually and not all the parts in a single download. If the scenario is scenario 35 or older, then the scenario is packaged in a single zip file, and this does not apply.
Increase downloading speed
The only ways to increase downloading speed are user-side. These include:
  1. Connecting to the fastest and most stable internet connection possible, ideally with an ethernet cable. 
  2. Using the internet connection when fewer people are using it, so your connection gets prioritized.
For reference, stable and fast connections tend to score over 50 Mbps (on the Ookla speed test) and download a 10 GB file in roughly 1 hour. Note that the download bottleneck and download time may change depending on the place around the world where the download is started.
Check if the file is corrupted
There are a few ways of checking if the file is corrupt. The easiest is to verify the file size. If the file is not its supposed file size, then the download was interrupted. The file size can be checked in the download link. For scenario 35 or older, go to File -> Info to see the download size. For more recent scenarios, the filesize is displayed with the file. 
More robust ways to check files include computing the MD5 hash. This is often unnecessary for our purposes, so we do not support this yet. 
If the problem persists
It should be noted that there is a luck component to a successful download over a sub-optimal connection. The download method used offers the best chances to transfer these large files. Please keep trying.
If the problem persists, contact one of our team members (João replies fast). We will figure out a workaround to get you the data you need.

Troubleshooting Dataset Extraction​

Watch a video explaining the steps of accessing the dataset

In more recent scenarios (scenario 36 and newer), we have split download data across multiple zip files. As scenarios get larger, this should facilitate the download. To understand how to extract these scenarios, watch the video to the right. If your scenario is older (scenario s <=35), this video may still be useful, but it is more likely that the solution to your problem lies further down below.

Understand the scenario structure

The final folder structure should be similar to the one on the right. The files are zipped in a way that, when extracted successfully, results in that structure. 

Important: Independently of the operative system, all methods require manually moving the scenario.csv file inside the extracted folder at the end of the extraction operation.


Steps to extract multi-part scenario zip files


1- Install the 7Zip utility (download here) to be able to extract ZIP files split in parts

2- Select all zip files downloaded, right-click, 7Zip, “Extract here”

Explanation: Each zip file has a folder called scenarioX, but different modality data inside. As we extract data from different modalities (e.g.,, etc), the result will be that the scenario folder from inside the zips of different modalities will merge automatically, leading to a single scenario folder with the right modality structure.

MacOS & Linux

Open the terminal.

Install 7zip (MacOS): brew install p7zip

Install 7zip (Ubuntu): sudo apt install p7zip

Navigate to the directory where all parts of the zip file are present. 

Extract using 7zip: 7z x

Here, should be replaced by the name of part 001 of the zip. Only this file needs to be included in the command.