Update project docs

README.md

@@ -1,17 +1,15 @@
 # Data processing
 
-**Work in proress.**
-
-Repository for data transformation workflow that processes BAR data from
-different sources and creates data dumps.
+Repository for [ETL](https://en.wikipedia.org/wiki/Extract%2C_transform%2C_load)
+workflow that processes BAR data from different sources and creates data dumps.
 
 At the moment the only functionality is a public dump of the past matches data
 combining information from teiserver and replays databases.
 
-## Access dumps
+## Data access
 
-Automatically generated documentation about accessing dumps generates from this
-repository is available at https://beyond-all-reason.github.io/data-processing/.
+Documentation about accessing dumps generated with pipeline in this repository
+is available at https://beyond-all-reason.github.io/data-processing/.
 
 ## Development
 

models/homepage.md

@@ -17,6 +17,21 @@ For example, to download [`matches`](#!/model/model.bar_data_processing.matches)
 model (data table) in Parquet format, URL is:
 https://data-marts.beyondallreason.dev/matches.parquet
 
+## Data quality
+
+The processing pipeline automatically removes some obviously very bad data (e.g.
+not clear to which team player belongs to), but still due to various bugs in
+teiserver, there are some issues with some small (low hundreds/tens) amount of
+entries in the dataset, for example:
+
+- Teiserver data reports different players vs teams assignments then intended
+- Some player skill is reduced as if it lost even though it won
+
+and likely plenty more.
+
+If someting should be filtered out as bad data: feel free to open a new issue in
+the repository.
+
 ## Schamas
 
 To browse schema of all exported models (data tables) navigate on the left side