This set of files contains a comprehensive example of estimating multinomial logit models in R using the mlogit package.
Simulated choice data are used so that the true model parameters are known. The simulated data are stored in the ./data folder. The following list provides a short description of the data files:
| File | Description |
|---|---|
data_mnl.csv |
Simulated choice data using a multinomial logit model. |
data_outsideGood.csv |
Similar data to the data_mnl.csv file, except this dataset include an "outside good" alternative (i.e. a "None" option in a conjoint survey). |
sensitivityCases.csv |
Cases used in the sensitivity analyses (see the 7.3 and 7.4 examples in the ./code folder). |
The .R files in the ./code folder illustrate different examples of models, plots, and analyses. The following list provides a short description of each file:
| File | Description |
|---|---|
0-simulateData.R |
Simulate the choice data. |
1.1-loadTools.R |
Load functions & libraries. |
1.2-exploreData.R |
Explore the data with summaries and plots. |
2.1-partworth_model.R |
Estimate a logit model with partworth parameters using mlogit. |
2.2-partworth_plots.R |
Plot the results of the partworth model using ggplot2. |
3.1-linear_model.R |
Estimate a logit model with linear parameters using mlogit. |
3.2-linear_plots.R |
Plot the results of the linear model using ggplot2. |
4.1-outsideGood_model.R |
Estimate a logit model with an outside good using mlogit. |
4.2-outsideGood_plots.R |
Plot the results of the outside good model using ggplot2. |
5-uncertainty.R |
Use the MASS library to take multivariate normal draws of the linear model coefficients and generate a 95% confidence interval of the coefficients. |
5.2-uncertainty_plots.R |
Plot the coefficients (with uncertainty) using ggplot2. |
6.1-wtp.R |
Compute the willingness to pay from the linear model (both point estimates and a 95% confidence interval using simulation). |
6.2-wtp_plots.R |
Plot the WTP results using ggplot2. |
6.3-wtp_logitr.R |
Directly estimate WTP using the logitr package. |
7.1-market_simulation.R |
Compute the expected market shares for a set of alternatives using coefficients from the linear model (both point estimates and 95% confidence intervals of the shares using simulation). |
7.2-marketSimulation_plots.R |
Plot the market simulation results using ggplot2. |
8.1-marketSensitivity.R |
Conduct a sensitivity analysis of the market shares to changes in attribute values. |
8.2-marketSensitivity_plots.R |
Plot results of the sensitivity analysis, including a 'tornado plot', using ggplot2. |
9.1-two_groups.R |
Full example analysis of two known groups that have distinct, different preferences. |
- Author: John Paul Helveston (www.jhelvy.com)
- Date First Written: Thursday, December 27, 2018
- License: GPL-3