Skip to content

reinvantveer/namari

Repository files navigation

Namari anomaly detector for QGIS

A data quality assessment plugin tool for QGIS. Fully automated anomaly or outlier detection over all properties of your data.

Data type support

Currently supports basically any QGIS data type, but some data is ignored:

Type Mapping
Feature Id Ignored
Text One-hot
Int32 Float
Int64 Float
Decimal Float
Date Seconds since 1970
DateTime Seconds since 1970
Boolean 0. or 1.
Binary Ignored

Note that each unique text value is assigned to its own attribute. This is because one-hot vectors are the conventional method for dealing with class-like data. If the text in your data is unique for each record, you'd better keep it out of the model builder, otherwise the expansion of the data will probably not result in a good data mapping.

Local dev-setup

This repo uses a combination of pipenv (unit testing) and Docker (integration testing) setups in order to develop and test. Use the following to set up an isolated virtual environment for development:

pipenv install --dev --deploy --site-packages --python=<path/to/your/QGIS/python/version> 

In order to obtain the path to your QGIS python interpreter, start QGIS, pres Ctrl-alt-P to start a python interpreter window and enter import sys; sys.executable

Troubleshooting

If you get something like

ImportError: /lib/x86_64-linux-gnu/libQt5DBus.so.5: undefined symbol: _ZN14QMetaCallEventC2EttPFvP7QObjectN11QMetaObject4CallEiPPvEPKS0_iiPiS5_P10QSemaphore, version Qt_5_PRIVATE_API

your pipenv environment wasn't initialized properly with qgis components. Follow the instructions from the local install setup using pipenv.