v1.1.2

✨ New features

• Added search radius parameter to IDW (#471)
• Added CLI function for feature importance and slightly modified the tool (#468)

🛠️ Other improvements

• Fixes to PCA tests (#465)

v1.1.1

🚀 Performance improvements

• Optimize distance computation (#455) (NOTE: This optimization affects also distance_to_anomaly, proximity_to_anomaly and proximity_computation)

🐞 Fixes

• Fix missing data closure requirement for CoDA transformations (#457)
• Fix plots not being saved correctly when also shown immediately in CLI functions (#461)

v1.1.0

🐞 Fixes

• Fix issue with winsorize transform when only one percentile was given (#450)
• Fix numeric input column checking for Local Moran's I tool (#453)

🛠️ Other improvements

• Pin scipy version to 1.11.4 to ensure Local Moran's I tool works (#451)

v1.0.3

✨ New features

• Min-max scaling can be used to scale data inversly (#436)
• Added Proximity computation tool (#432)
• Added Proximity to anomaly tool (#440)
• Added Agterberg-Cheng CI test for Weights of evidence (#343)
• Added optimized version of Distance to anomaly tool (requires Windows and to ensure installation of gdal_array) (#423)

🐞 Fixes

• Fixed transformations tool failing when raster metadata did not have nodata defined (#436)
• Fixed PCA tools and updated outputs (#446)

🛠️ Other improvements

• Modified/improved Weights of evidence tool (#355)
• Removed defaults channel from conda installation instructions (#434)
• Added issue templates for new tool requests and bug reports (#317)

v1.0.2

✨ New features

• Added CLI functions for MLP classifier and regessor (#412)
• Added Mask raster tool (#413)
• Added CLI function for Mask raster tool (#419)

🐞 Fixes

• Fixed Keras model saving and available metrics (#412)
• Fixed return data dimension when predicting with Keras regressor model (#412)
• Fixed issues with CoDa tools (#416 )

🛠️ Other improvements

• Added masking parameter to Unify raster grids tool (#417)
• Added band parameter to Descriptive statistics raster too (#428)

v1.0.1

🐞 Fixes

• Fix unify raster grids copying unwanted raster metadata (#402)
• Fix invalid scoring for binary classification models (#408)

🛠️ Other improvements

• Reduce number of shown decimals in numeric tool outputs (#411)
• Extend documentation (Conda installation) (#400)

v1.0.0

✨ Available tools

Conversions

• CSV to geodataframe
• Raster to dataframe

Evaluation

• Calculate base metrics (true positive rate, false positive rate, proportion of area)
• Summarize label metrics (for binary classification)
• Summarize probability metrics (for binary classification)
• Plot ROC (receiver operating characteristic) curve
• Plot DET (detection error tradeoff) curve
• Plot precision-recall curve
• Plot calibration curve (reliability curve)
• Plot distribution of predicted probabilities
• Plot confusion matrix
• Plot neural network loss
• Plot neural network accuracy
• Plot prediction area curves
• Plot rate curve
• Score predictions (mae, mse, rmse, r2, accuracy, precision, recall, f1)

Exploratory analysis

• Basic plots (re-exports from Seaborn – barplot, boxplot, ecdfplot, heatmap, histplot, kdeplot, lineplot, pairplot, regplot, scatterplot)
• Chi-square test
• Correlation matrix & plot correlation matrix
• Covariance matrix
• DBSCAN (array & dataframe versions)
• Descriptive statistics (raster & geodataframe versions)
• Evaluate feature importance (for ML model)
• K-means clustering (array & dataframe versions)
• Local Moran's I
• Normality test (array & dataframe versions)
• Plot parallel coordinates
• Compute PCA & plot PCA

Prediction

• Fuzzy overlay (AND, OR, PRODUCT, SUM, GAMMA)
• Train gradient boosting classifier & regressor
• Train random forest classifier & regressor
• Train logistic regression model
• Train MLP classifier & regressor
• ML modeling utility tools: save model, load model, split data, reshape predictions, prepare data for ML, read data for evaluation
• Predict classifier & regressor

Raster processing

• Clip raster
• Create constant raster
• Distance to anomaly
• Extract values from raster
• Reclassify (manual breaks, defined intervals, equal intervals, quantiles, natural breaks, geometric intervals, standard deviation)
• Reproject
• Resample
• Snap
• Unify rasters
• Unique combinations
• Extract window
• Filters
  • 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
• Derivatives
  • First order surface derivatives
  • Second order surface derivatives
  • Classify aspect

Training data tools

• Balance classes (SMOTETomek)

Transformations

• Binarize
• Clip transform
• Min-max scale
• Z-score normalize
• Log transform (ln, log2, log10)
• One-hot encode
• Sigmoid transform
• Winsorize
• CODA transforms
  • ALR transform
  • Inverse ALR transform
  • CLR transform
  • Inverse CLR transform
  • Single ILR transform
  • Single pairwise logratio
  • Pairwise logratio
  • Single PLR transform
  • PLR transform

Vector processing

• Calculate geometry (length for line, area for polygon)
• Cell based association
• Distance computation
• Extract shared lines
• IDW interpolation
• Kriging interpolation
• Rasterize vector
• Reproject vector
• Vector density

Utilities

• Various utilities regrading rasters, vectors, nodata and more. Individual tools/functions not listed here

🌱 Work in progress

• Weights of evidence (calculate weights and calculate responses) – this tool is already included in EIS Toolkit but will undergo some changes in near future
• CNN classifier & regressor models
• Autoencoder for image segmentation
• Mini-Unet for image segmentation
• Data sampler
• Mahalanobis similarity
• Bayesian NN

v0.5.2

Minor fix release 2 for EIS Toolkit 1.0 release candidate

✨ New features

• Added CLI tool for CBA (#391)

🐞 Fixes

• Minor fixes/modifications for CBA (#391)

🛠️ Other improvements

• Enabled JSON str outputs in CLI again (#394)

v0.5.1

Minor fix release for EIS Toolkit 1.0 release candidate

✨ New features

• Added CLI tools for evaluation tools (Summarize label metrics binary, Summarize probability metrics, Plot ROC curve, Plot DET curve, Plot precision recall curve, Plot calibration curve, Plot predicted probability distribution and Score predictions) (#387)

🐞 Fixes

• Fix result prints for some CLI functions (#390)
• Fix nodata masking for ML models run through CLI

🛠️ Other improvements

• Separated CLI functions for classifier and regressor predicting and testing (prev. called evaluate) (#390)
• Added classification threshold parameter for predicting with a classifier (#390)

v0.5.0

EIS Toolkit 1.0 release candidate

This release is the final beta release which serves as a release candidate for the 1.0 release that will be done soon. Bug fixes and minor adjustments may be still made before the 1.0 release.

✨ New features

• Add Distance to anomaly (#324)
• Add K-means clustering for raster data (#365)
• Add DBSCAN for raster data (#366)
• Add Confusion matrix plot (#363)
• Add Neural network accuracy and loss plots (#362)
• Add some raster utility tools (#371)
• Add many evaluation tools: Summarize label metrics binary, Summarize probability metrics, Plot ROC curve, Plot DET curve, Plot precision recall curve, Plot calibration curve, Plot predicted probability distribution and Score predictions (#368)
• Add CLI implementations for many tools

🐞 Fixes

• Fix rasterization of vector label data for ML models called from CLI
• Fix fuzzy sum logic (#381)

🛠️ Other improvements

• Add Criterion parameter to Random Forest models (#345)
• Add nodata handling to Raster to dataframe tool (#364)
• Add max distance parameter to Distance computation and Distance to anomaly (#382)
• Change resampling_method parameter type in Resample raster tool to literal/str (#349)
• Refactor vector -> raster tools: base raster and pixel size + extent to define output raster grid are now available for all of the tools (#375 and #376)
• Refactor validation category. Now called evaluation (#368)