Open qszhao opened 1 month ago
This is an ongoing process to determine how and whether to use the validation data created by @Adenikemie and @AlexandraMiddleton as part of the creation of new training data.
Progress so far is that a meeting was held with the above + @manuellacomerio - outcome: reporting to @qszhao all of the workflows that they have done.
Next steps - a subsequent meeting to determine information about the grid that was used and to see whether the grid matches.
@manuellacomerio will convert the SDS/ISL output grid for Lagos and Kano from the output files and try to match it with Adenike/Alex's validation data. If the grid matches, the data will be joined. She will also try to match it with Nairobi data Monika supplied.
Then PAR data will be joined to the grid. @andymithamclarke if you can help @manuellacomerio to identify the PAR results that will be great. Now it is a bit confused which one we should use.
Closing this as complete - see update in #101
Reopen the issue - we will need to generate the new high/medium/low categories for morphological informality based on the reference data in Nairobi and Lagos, plus the PAR results for SDS and ISL.
Reference dataset with L/M/H is ready for Nairobi. This has been shared over CRIB and GDrive.
Reference dataset with L/M/H are planned for Friday 18. @SebastianHafner @manuellacomerio will update on the progress
@SebastianHafner to upload reference to GDrive and CRIB for Lagos. @SebastianHafner to close this issue once all the data is uploaded to the appropriate locations.
We will finalise the validation datasets: