forked from NorskRegnesentral/shapr
-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathREADME.Rmd
139 lines (101 loc) · 6.38 KB
/
README.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
---
output: github_document
bibliography: ./inst/REFERENCES.bib
---
<!-- README.md is generated from README.Rmd. Please edit that file -->
```{r setup, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "100%",
tidy = "styler"
)
```
# shapr <img src="man/figures/NR-logo_utvidet_r32g60b136_small.png" align="right" height="50px"/>
<!-- badges: start -->
[](https://circleci.com/gh/NorskRegnesentral/shapr)
[](https://github.com/NorskRegnesentral/shapr/actions?query=workflow%3AR-CMD-check)
[](https://www.tidyverse.org/lifecycle/#experimental)
[](https://opensource.org/licenses/MIT)
<!-- badges: end -->
The most common task of machine learning is to train a model which is able to predict an unknown outcome (response variable) based on a set of known input variables/features.
When using such models for real life applications, it is often crucial to understand why a certain set of features lead to exactly that prediction.
However, explaining predictions from complex, or seemingly simple, machine learning models is a practical and ethical question, as well as a legal issue. Can I trust the model? Is it biased? Can I explain it to others? We want to explain individual predictions from a complex machine learning model by learning simple, interpretable explanations.
Shapley values is the only prediction explanation framework with a solid theoretical foundation (@lundberg2017unified). Unless the true distribution of the features are known, and there are less than say 10-15 features, these Shapley values needs to be estimated/approximated.
Popular methods like Shapley Sampling Values (@vstrumbelj2014explaining), SHAP/Kernel SHAP (@lundberg2017unified), and to some extent TreeSHAP (@lundberg2018consistent), assume that the features are independent when approximating the Shapley values for prediction explanation. This may lead to very inaccurate Shapley values, and consequently wrong interpretations of the predictions. @aas2019explaining extends and improves the Kernel SHAP method of @lundberg2017unified to account for the dependence between the features, resulting in significantly more accurate approximations to the Shapley values.
[See the paper for details](https://arxiv.org/abs/1903.10464).
This package implements the methodology of @aas2019explaining.
The following methodology/features are currently implemented:
- Native support of explanation of predictions from models fitted with the following functions
`stats::glm`, `stats::lm`,`ranger::ranger`, `xgboost::xgboost`/`xgboost::xgb.train` and `mgcv::gam`.
- Accounting for feature dependence assuming the features are Gaussian (@aas2019explaining).
- Accounting for feature dependence with a Gaussian copula (Gaussian dependence structure, any marginal) (@aas2019explaining).
- Accounting for feature dependence using the Mahalanobis distance based empirical (conditional) distribution approach of @aas2019explaining.
- Combine any of the three methods.
- Optional use of the AICc criterion of @hurvich1998smoothing when optimizing the bandwidth parameter in the empirical (conditional) approach of @aas2019explaining.
- Functionality for visualizing the explanations.
- Support for models not supported natively.
<!--
Current methodological restrictions:
- The features must follow a continuous distribution
- Discrete features typically work just fine in practice although the theory breaks down
- Ordered/unordered categorical features are not supported
-->
Future releases will include:
- Support for parallelization over explanations, Monte Carlo sampling and features subsets for non-parallelizable prediction functions.
- Computational improvement of the AICc optimization approach
- Adaptive selection of method to account for the feature dependence
Note that both the features and the prediction must be numeric. The approach is constructed for continuous features. Discrete features may also work just fine with the empirical (conditional) distribution approach.
Unlike SHAP and TreeSHAP, we decompose probability predictions directly to ease the interpretability, i.e. not via log odds transformations.
The application programming interface (API) of `shapr` is inspired by @lime_api.
## Installation
To install the current development version, use
```{r, eval = FALSE}
devtools::install_github("NorskRegnesentral/shapr")
```
## Example
`shapr` supports computation of Shapley values with any predictive model which takes a set of numeric features and produces a numeric outcome.
The following example shows how a simple `xgboost` model is trained using the *Boston Housing Data*, and how `shapr` explains the individual predictions.
```{r basic_example, warning = FALSE}
library(xgboost)
library(shapr)
data("Boston", package = "MASS")
x_var <- c("lstat", "rm", "dis", "indus")
y_var <- "medv"
x_train <- as.matrix(Boston[-(1:6), x_var])
y_train <- Boston[-(1:6), y_var]
x_test <- as.matrix(Boston[1:6, x_var])
# Looking at the dependence between the features
cor(x_train)
# Fitting a basic xgboost model to the training data
model <- xgboost(
data = x_train,
label = y_train,
nround = 20,
verbose = FALSE
)
# Prepare the data for explanation
explainer <- shapr(x_train, model)
# Specifying the phi_0, i.e. the expected prediction without any features
p <- mean(y_train)
# Computing the actual Shapley values with kernelSHAP accounting for feature dependence using
# the empirical (conditional) distribution approach with bandwidth parameter sigma = 0.1 (default)
explanation <- explain(
x_test,
approach = "empirical",
explainer = explainer,
prediction_zero = p
)
# Printing the Shapley values for the test data
print(explanation$dt)
# Finally we plot the resulting explanations
plot(explanation)
```
## Contribution
All feedback and suggestions are very welcome. Details on how to contribute can be found
[here](./.github/CONTRIBUTING.md)
Please note that the 'shapr' project is released with a
[Contributor Code of Conduct](CODE_OF_CONDUCT.md).
By contributing to this project, you agree to abide by its terms.
## References