gispocoding/eis_qgis_plugin

A QGIS plugin for mineral prospectivity mapping

eis_qgis_plugin's Issues

Add Fuzzy overlay to modeling wizard

Implement a density scatterplot with pyqtrgraph

This density scatterplot should be interactive and allow the user to select some data area, and the selection should be reflected on the map. 3D would be preferable, but 2D prototype will do too. The desired functionality is close to an old QGIS plugin Raster Data Plotting. This feature relates to issue #8 .

Fix Unify rasters alg

Update PCA processing algorithm

Divide PCA processing algorithm into raster and vector variants to match the EIS Toolkit CLI definition

Remove parameter in kriging processing algorithm

Add processing algorithms for random forest, gradient boosting and logistic regression

Tasks

Beta Give feedback

1. Logistic regression
2. Random forest classifier
3. Random forest regressor
4. Gradient boosting classifier
5. Gradient boosting regressor

Update README

Tasks

Beta Give feedback

1. Update description about the plan
2. Update architecture diagram
3. Add testing instructions and requirements
4. Update or remove development docs

Prototype dockable version of EIS Wizard

Extend configuration file for processing algorithms

Each processing algorithm needs information from configuration file to be initialized.

It is still unclear if TOML files can be used, as it is not default Python package. However, it seems to be present in QGIS Python in OS and Windows, and could be installed via script for Linux users.

If TOML files cannot be used, INI will likely be used instead.

Implement EIS Wizard with vertical list menu

Add Reclassify raster processing algorithm

Note: CLI function for Reclassify raster should be first implemented In EIS Toolkit.

Add functionality to process button

Design UI for basic seaborn plots

Check licensing

It should be checked if EIS QGIS Plugin (and QGIS plugins overall) can be licensed under the EUPL-1.2.

Add option to use custom/no mineral system and/or scale in proxy selection

Add Coda transformations algs

Add MLP to modeling wizard

Write logic to summary statistics (EIS Explore Data summary tab)

Design model data preparation view

A need for model data preparation was identified. This data preparation was seen to be different from producing proxies, and was not wanted (at least for now) to combined in the same view with modeling, and therefore a new view needs to be designed. This view could facilitate at least the following processes:

• Normalize data
• Replace nodata
• Encode data
• Transform data
• Split data
• Generate data
• Balance data

Add color coding for importance and/or keywords in EIS Preprocess window

Add Extract shared lines and calculate geometry algs

Improve and develop CLI for plugin - toolkit communication

The actual communication works at the moment using subprocess.Popen to call an external script (Python with EIS Toolkit). What remains to be done is to decide whether information is delivered:

• Using a configuration file, such as JSON. The subprocess would only deliver path to the configuration file the plugin has just written
• Using a CLI library, such as Click or Typer, to define the ways of argument passing
• Delivering arguments via the command line "raw" and with custom parsing at the Toolkit end

Add processing algorithms for IDW and kriging

Design and implement prototype model view

The model view could be a wizard or one dialog (if everything fits into one view). In model view, the following features need to be available:

• Select model
• Select data (and description of needed data)
• Select model parameters (with some presets available)
• Model report / validation
• (Optimization methods for model parameter selection)
• (Guidance)

Add pre-commit hooks

Add processing algorithms for MLP

Tasks

Beta Give feedback

1. MLP classifier
2. MLP regressor

Add Weights of evidence to modeling wizard

Design how the EIS Wizard will communicate with the Toolkit

At the moment, a likely alternative is that the Wizard will call EIS processing algorithms, which are the only QGIS entities communicating with the Toolkit. This simplifies the relation of Toolkit - Wizard - Processing, but could restrict computation efficiency and forming of complex workflows in some cases.

Add Feature importance processing algorithm

Note: CLI function for Feature importance should be first implemented In EIS Toolkit.

Add tabs to Modeling section in Wizard

Add the following tabs to Modeling:

• Model data preparation
• Model creation
• Model application
• Model evaluation

Design and implement prototype one proxy creation view

This first proxy creation interface could allow the user the execute the following common workflow:

• Select a vector file and relevant attribute(s) from the file/layer
• Select which values are relevant (QgsExpression?)
• Calculate (euclidean) distance, in other words rasterize the vector in a specific way

This proxy creation opens when "Process" button is clicked in the proxy/preprocessing table view.

Add Surface derivatives processing algorithm

Include Seaborn in the plugin

If Seaborn could be baked into the plugin as an external library, users would not have to manually install it. It should be investigated if this is possibly and what are the potential drawbacks.

Add Distance to anomaly processing algorithm

Note: CLI function for Distance to anomaly should be first implemented In EIS Toolkit.

Add processing algorithms

Add tools from EIS Toolkit that are already implemented as processing algorithms.

The following algorithms should be added:

• Distance computation

• Vector density

• Reproject vector

• DBSCAN

• Kmeans cluster

• Extract values from raster

• Windowing

• Calculate AUC

• Calculate base validation metrics

• Plot prediction area curve (modifications to the plotting might be done in the future)

• Plot rate curve (modifications to the plotting might be done in the future)

• Rename statistics report

If there are problems or skeletons with wrong input and output in some existing implementations, they should be changed too.

Add proxies

Add proxies and all related information to the proxy setup file. The needed information is at least:

• Proxy name
• Related mineral system type
• Importances per scale
• Workflow steps needed for calculation / UI name that allows for the production of the proxy
• Keywords / categories

Add bivariate analysis tab to explore

Customize icons for menu items

Some of the icons currently used for menu items (Mineral system proxies, EDA, Modelling, Settings and About should be modified.

Design and implement prototype explore view

The explore view is part of the wizard that facilitates EDA (exploratory data analysis). It should enable producing plots and to use some data analysis methods such as clustering and PCA. At the least the following plots are required:

• Scatterplot (vector data)
• Histogram (vector data and raster data)
• Line plot (vector data)
• Box plot (vector data)
• Parallel coordinates (vector data)
• PCA visualization (2D / 3D ROI (region of interest) type plot, scatterplot matrix etc.)

Plots that are needed only for validation do not need to be available here.

Implement alpha version of the Wizard

Alpha version of the Wizard should be able to communicate with the Toolkit and how some working interface with realistic elements for MPM. It does not need to be useful for actual work or have a carefully designed UI.

Add Local Moran's I processing algorithm

Note: CLI function for Local Moran's I should be first implemented In EIS Toolkit

Add processing algorithms for filter/smoothing functions

• Focal filter
• Gaussian filter
• Mexican hat filter
• Lee additive noise filter
• Lee multiplicative noise filter
• Lee additive multiplicative noise filter
• Lee enhanced filter
• Gamma filter
• Frost filter
• Kuan filter

Add study scale/mineral system selection dialog

Add UI drafts for the 4 proxy preparation workflows

Tasks

Beta Give feedback

1. Workflow 1
2. Workflow 2
3. Workflow 3
4. Workflow 4

Redesign plotting interfaces in Wizard

I'll try if one common dialog for all plots would make sense. The first parameter would be plot type and that would dictate which parameters are displayed. The produced plot could be put in another page of a stacked widget, since on the side of parameters it gets easily squished or doesn't have enough space.

Add Cell-based association processing algorithm

Note: CLI function for CBA should be first implemented In EIS Toolkit.

Add clipping / area selection parameter to modeling wizard

A requested feature for the modeling wizard is a vector/extent input that allows selecting an area to use for training a machine learning model. If this feature was implemented, the user would not need to clip each of their datasets individually to produce training and testing tests, but instead could specify the areas on the fly.

Implementing this needs some thought how the clipping/masking is best handled. Some options are:

• When the user presses "Start training", clip_raster and clip_vector processing algorithms are executed for each input dataset and these will be used to train the model
• The CLI function receives the (optional) extent/polygon for clipping as a parameter and clipping/masking is done in the CLI function
• The clip/mask parameter goes all the way to the EIS Toolkit training function (which needs a new corresponding parameter) which will mask irrelevant data out

Fix resample processing alg

Add produced proxy layers as a tree/layer tree widget in proxy selection table

Produced proxy layers should be listed under the proxy name. The tree widget/view (QTreeView) or layer tree view (QgsLayerTreeView) could be good choices for this job. Additionally, a functionality to select final version among produced layers for a certain proxy has been requested.

Design modeling view in EIS Wizard