Finish new heatmap schema

[EPajares]

Goal of this issue

For each calculation types for heat-maps (#1783), we need different settings. For example for modified_gaussian_per_grid(), we need sensitivity and cutoff. We need to check these settings.

Resources

Examples of configs: Config for gravity:

{
   "supermarket":{
      "max_traveltime":10,
      "sensitivity":250000,
      "weight":1
   },
   "tram_stop":{
      "max_traveltime":15,
      "sensitivity":250000,
      "weight":1
   }
}

Config for closest:

{
   "supermarket":{
      "max_traveltime":5,
      "weight":1
   },
   "tram_stop":{
      "max_traveltime":10,
      "weight":1
   }
}

Deliverables

Branch to derive

[metemaddar]

I think the config is better to be like:

[
    {
      "category": "supermarket",
      "max_traveltime":10,
      "sensitivity":250000,
      "weight":1
   },
   {
      "category": "tram_stop",
      "max_traveltime":15,
      "sensitivity":250000,
      "weight":1
   }
]

Then we can use a general Config schema for them. And validate based on term category.

[metemaddar]

Of course this can have a con that the input can have multiple values for single category :thinking:

[EPajares]

Hard decision. I think your idea makes sense. But maybe we could make sure that the category is only used once. So we avoid that the user defines two configs for one category.

[metemaddar]

So we can create a Pydantic model per heat-map type (closest, gravity, etc.) and based on type, validate each configurations against it's related Pydantic class.

[EPajares]

Sounds great

[metemaddar]

Implemented (Initiated) for gravity and closest: 5c9e5cec01b541d384ae1188104294ed1f8f3a8e We need to add others as well.

[metemaddar]

• [ ] :warning: We need to check this configurations as the configuration style changed.

https://github.com/goat-community/goat/blob/9c05fa5c6b68357ed2af591190bcc8b25c840d8c/app/api/src/crud/crud_heatmap.py#L678

[metemaddar]

I think it's also good to use some type key for poi/aoi/... This approach makes code cleaner for validation.

{
   "supermarket":{
      "max_traveltime": 5,
      "weight": 1,
      "type": "aoi"
   },
   "park":{
      "max_traveltime": 10,
      "weight": 1
      "type": "poi"
   }
}

[metemaddar]

Of course this also has cons 🤔 As we may have different pydantic class for poi or aoi But we can also make fields optional and then do validation based on type.

[EPajares]

I agree. Or should we make poi/aoi as parent key like:

{ "aoi": {} "poi": {} }

[p4b-bro[bot]]

This task/issue closed on Tue Jun 06 2023 ✅