Datasets:

Modalities:
Tabular
Text
Formats:
parquet
Languages:
English
Tags:
legal
Libraries:
Datasets
pandas
License:
dieineb commited on
Commit
6317532
·
verified ·
1 Parent(s): add8d05

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +11 -6
README.md CHANGED
@@ -99,16 +99,23 @@ size_categories:
99
 
100
  ## Overview
101
 
102
- The COMPAS (Correctional Offender Management Profiling for Alternative Sanctions) dataset is a popular commercial
103
- algorithm used by judges and parole officers for scoring criminal defendant’s likelihood of reoffending (recidivism).
104
 
105
 
106
  ## Dataset Details
107
 
108
  This data originally came from [COMPAS Recidivism Racial Bias](https://www.kaggle.com/datasets/danofer/compass)
109
 
110
- The original dataset contains :
 
 
 
111
 
 
 
 
 
112
 
113
  ```latex
114
  @incollection{angwin2022machine,
@@ -122,14 +129,12 @@ The original dataset contains :
122
  ```
123
  - Dataset Name: **COMPAS**
124
  - Language: English
125
- - Total Size:
126
 
127
  ## Contents
128
  Data contains variables used by the COMPAS algorithm in scoring defendants and their outcomes within 2 years of the decision for over 10,000 criminal defendants in Broward County, Florida.
129
  3 subsets of the data are provided, including a subset of only violent recidivism (as opposed to, e.g. being reincarcerated for non-violent offenses such as vagrancy or Marijuana).
130
 
131
- An in-depth analysis by ProPublica can be found in their [data methodology article](https://www.propublica.org/article/how-we-analyzed-the-compas-recidivism-algorithm).
132
-
133
 
134
  ## How to use
135
 
 
99
 
100
  ## Overview
101
 
102
+ The COMPAS (Correctional Offender Management Profiling for Alternative Sanctions) dataset has been used to analyze various aspects of the criminal justice system, including whether certain groups of defendants are more likely
103
+ to receive harsher sentences, whether the algorithm's predictions are accurate, and the impact of these predictions on sentencing decisions. However, it's crucial to consider the context and potential biases when interpreting the results from this dataset.
104
 
105
 
106
  ## Dataset Details
107
 
108
  This data originally came from [COMPAS Recidivism Racial Bias](https://www.kaggle.com/datasets/danofer/compass)
109
 
110
+ The 40 columns provide a detailed look at the defendant's demographic information, assessment details,
111
+ recidivism indicators, and risk scores assigned by the COMPAS algorithm. Each row represents a unique assessment or case,
112
+ capturing the information associated with an individual's involvement with the criminal justice system and the assessment made by the
113
+ COMPAS tool regarding their risk of Recidivism and other factors.
114
 
115
+ An in-depth analysis by ProPublica can be found in their [data methodology article](https://www.propublica.org/article/how-we-analyzed-the-compas-recidivism-algorithm).
116
+
117
+ Original Article:
118
+ [Machine Bias](https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing)
119
 
120
  ```latex
121
  @incollection{angwin2022machine,
 
129
  ```
130
  - Dataset Name: **COMPAS**
131
  - Language: English
132
+ - Total Size: 18,316 demonstrations
133
 
134
  ## Contents
135
  Data contains variables used by the COMPAS algorithm in scoring defendants and their outcomes within 2 years of the decision for over 10,000 criminal defendants in Broward County, Florida.
136
  3 subsets of the data are provided, including a subset of only violent recidivism (as opposed to, e.g. being reincarcerated for non-violent offenses such as vagrancy or Marijuana).
137
 
 
 
138
 
139
  ## How to use
140