The Open Repair Alliance has just released its latest set of repair data logged from community repair events around the world - this is the 6th aggregation, and now the dataset is over 81,000 records, a big increase from last time.
The data that we all enter in to Restarters.net goes in to this combined data set, along with data from Repair Cafe International, anstiftung, Fixit Clinic and Repair Cafe Wales - and newly added this time, Repair Cafe Denmark.
You can see on the Insights pages at openrepair.org how this data is analysed and fed in to policy discussions.
This is excellent News,
ReparatorAI will jump into it,
Any new features in the dataset?
FYI , to feed our ReparatorAI tool we go through a series of enhancement actions on tje dataset:
We clean existing data (brand names typos, fill in empty values)
We add a superObject class attribute for easy navigation (kitchen, bathroom, office, …) between the long list of product categories
We have used the “problem description column” to classify between different type of defects (electrical, mechanical, fire/dust/water, general, unknown) after translation of the whole languages
We think this brings a lot of value to the dataset,
We will apply these steps to the new delivery with 81000 repairs (+30% !)
Would.you be interested in discussing this with us?
Always great to hear what you are building on top of the data with ReparatorAI
No changes to the structure - it is still ORDS v0.3 like last time. Just more data!
We try to do cleaning upstream as well as much as is possible, so if there’s a way you can report back which data points you need to clean, we may be able to resolve them at source.
Interesting, is this classification with natural language processing?
