# logistic_regression¶

This module implements logistic regression for scenario discovery.

The module draws its inspiration from Quinn et al (2018) 10.1029/2018WR022743 and Lamontagne et al (2019). The implementation here generalizes their work and embeds it in a more typical scenario discovery workflow with a posteriori selection of the appropriate number of dimensions to include. It is modeled as much as possible on the api used for PRIM and CART.

class ema_workbench.analysis.logistic_regression.Logit(x, y, threshold=0.95)

Implements an interactive version of logistic regression using BIC based forward selection

Parameters: x (DataFrame) – y (numpy Array) – threshold (float) –
coverage

coverage of currently selected model

Type: float
density

density of currently selected model

Type: float
res_dim

number of restricted dimensions of currently selected model

Type: int
peeling_trajectory

stats for each model in peeling trajectory

Type: DataFrame
models

list of models associated with each model on the peeling trajectory

Type: list
inspect(i, step=0.1)

Inspect one of the models by showing the threshold tradeoff and summary2

Parameters: i (int) – step (float between [0, 1]) –
plot_pairwise_scatter(i, threshold=0.95)

plot pairwise scatter plot of data points, with contours as background

Parameters: i (int) – threshold (float) – Figure instance

The lower triangle background is a binary contour based on the specified threshold. All axis not shown are set to a default value in the middle of their range

The upper triangle shows a contour map with the conditional probability, again setting all non shown dimensions to a default value in the middle of their range.

run()

run logistic regression using forward selection using a Bayesian Information Criterion for selecting whether and if so which dimension to add

show_threshold_tradeoff(i, cmap=<matplotlib.colors.ListedColormap object>, step=0.1)

Visualize the trade off between coverage and density for a given model i across the range of threshold values

Parameters: i (int) – cmap (valid matplotlib colormap) – step (float, optional) – a Figure instance
show_tradeoff(cmap=<matplotlib.colors.ListedColormap object>)

Visualize the trade off between coverage and density. Color is used to denote the number of restricted dimensions.

Parameters: cmap (valid matplotlib colormap) – a Figure instance
update(model, selected)

helper function for adding a model to the colection of models and update the associated attributes

Parameters: model (statsmodel fitted logit model) – selected (list of str) –