ChoiceModels.jl — A symbolic, extensible package for estimating Discrete Choice Models in Julia

ChoiceModels.jl is a Julia package designed for flexible and symbolic specification of Discrete Choice Models (DCM). Inspired by Biogeme and Apollo, it uses symbolic algebra, modular architecture, and Julia's type system to allow rapid prototyping and estimation of models like Logit and Mixed Logit.

---

🚀 Key Features

• Symbolic utility expressions using Parameter, Variable, and algebraic operators.
• Multinomial Logit and Mixed Logit estimation with availability conditions.
• Integration with DataFrames.jl for data handling.
• Estimation via Optim.jl, supporting automatic or analytic gradients.
• Prediction tools for probabilities and most likely alternatives.
• Extensibility for new models (e.g., Latent Class, Nested Logit).

---

✨ Quick Example: Estimating a Logit Model

using ChoiceModels, CSV, DataFrames

df = CSV.read("my_data.csv", DataFrame)

asc_car = Parameter(:asc_car, value=0.0)
β_time = Parameter(:β_time, value=0.0)
β_cost = Parameter(:β_cost, value=0.0)

V_car = asc_car + β_time * Variable(:time_car) + β_cost * Variable(:cost_car)
V_bus = β_time * Variable(:time_bus) + β_cost * Variable(:cost_bus)

model = LogitModel([V_car, V_bus]; data=df, availability=[trues(nrow(df)), trues(nrow(df))])
results = estimate(model, :choice)
probs = predict(model, results)

---

🌀 Example: Estimating a Mixed Logit Model

using ChoiceModels, CSV, DataFrames

df = CSV.read("my_data.csv", DataFrame)

asc_car = Parameter(:asc_car, value=0.0)
β_time = Parameter(:β_time, value=0.0)
σ_time = Parameter(:σ_time, value=1.0)

draw = Draw(:time_rnd)
V_car = asc_car + (β_time + σ_time * draw) * Variable(:time_car)
V_bus = (β_time + σ_time * draw) * Variable(:time_bus)

model = MixedLogitModel([V_car, V_bus];
    data=df,
    idvar=:id,
    R=500,
    draw_scheme=:mlhs,
    availability=[trues(nrow(df)), trues(nrow(df))])

results = estimate(model, :choice)
P = predict(model, results)

---

📦 Installation

] dev https://github.com/ighdez/ChoiceModels.jl

---

🧱 Architecture Overview

• Expressions: Parameter, Variable, Draw, and algebraic combinators
• Models: LogitModel, MixedLogitModel (more to come!)
• Estimation: estimate(model, choices) using Optim.jl
• Prediction: predict(model, results) returns probabilities

---

📚 Documentation

All public functions are documented via Julia docstrings. Use ?estimate in the REPL to see inline help. Full documentation will be published using Documenter.jl soon.

---

📄 License

MIT License

---

🙌 Acknowledgements

• Biogeme
• Apollo
• Discrete Choice Methods with Simulation

Built with ♥ and ChatGPT pair-programming.