Data-driven causal-loop and feedback network analysis in R.
CausalLoopAnalytics provides a compact workflow for signed causal models: construct a causal-loop diagram (CLD), discover temporal edges with a Granger-style test, detect and classify reinforcing/balancing feedback loops, quantify loop strength, assess bootstrap stability, calculate network centrality and leverage scores, compare models, and explore structural interventions.
The package is intentionally focused on evidence and interpretation of feedback structure, rather than network enumeration or stochastic differential-equation simulation. Causal discovery output retains polarity, effect weight, p-value and method metadata.
library(CausalLoopAnalytics)
m <- create_cld(c("Antibiotic use", "AMR", "Treatment failure"))
m <- add_link(m, "Antibiotic use", "AMR", "+", 0.8)
m <- add_link(m, "AMR", "Treatment failure", "+", 0.7)
m <- add_link(m, "Treatment failure", "Antibiotic use", "+", 0.6)
find_feedback_loops(m)
classify_loops(m)
loop_strength(m)
loop_stability(m, B = 100, seed = 1)
network_summary(m)
identify_leverage_points(m)
plot_cld(m)
plot_loops(m)model <- infer_causal_network(my_time_series, lag = 2, alpha = 0.05)
summary_causal_loop(model)The Granger-style routine identifies temporal predictive relationships. It does not establish mechanistic causation and should be interpreted alongside study design and domain knowledge.
install.packages("CausalLoopAnalytics_1.0.0.tar.gz", repos = NULL, type = "source")The methods are domain-agnostic and can be used for human medicine, veterinary medicine, agriculture, epidemiology, ecology, public health and One Health systems.