Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/smileexpression/los-angeles-crime

This experiment aims to predict crime in Los Angeles through machine learning methods.
https://github.com/smileexpression/los-angeles-crime

ipynb logistic-regression machine-learning mlp python

Last synced: 1 day ago
JSON representation

This experiment aims to predict crime in Los Angeles through machine learning methods.

Awesome Lists containing this project

README 

        
# Los-Angeles-Crime



机器学习实验课大作业



## 环境



Python 3.9.17



## 数据集



[Los Angeles Crime Data 2020-2023](https://www.kaggle.com/datasets/asaniczka/crimes-in-los-angeles-2020-2023/data)



特征



> "division_number","date_reported","date_occurred","area","area_name","reporting_district","part","crime_code","crime_description","modus_operandi","victim_age","victim_sex","victim_descent","premise_code","premise_description","weapon_code","weapon_description","status","status_description","crime_code_1","crime_code_2","crime_code_3","crime_code_4","location","cross_street","latitude","longitude"



“部门编号”、“报告日期”、“发生日期”、“区域”、“区域名称”、“报告地区”、“部分”、“犯罪代码”、“犯罪描述”、“作案方式”、“受害者年龄”、“受害者性别”、“受害者血统”、"前提代码"、"前提描述"、"武器代码"、"武器描述"、"状态"、"状态描述"、"犯罪代码1"、"犯罪代码2"、"犯罪代码3"、"犯罪代码4"、"位置"、"交叉街道"、"纬度 “，“经度”



## 分工



1. 导包、导入数据集（照旧）



2. clean data



   - 这部分综合algorithms/data_clean.py和你MLTTest中对数据集的处理方式



   - 删除不必要的信息，这几个是一定没有用的（至于这两个删除的顺序，训练结果差别应该不大）



     ```python

     # 删除没有用的列

     data.drop(

         [

             "division_number",

             "date_reported",

             "area_name",

             "reporting_district",

             "part",

             "crime_description",

             "modus_operandi",

             "premise_description",

             "weapon_description",

             "status_description",

             "crime_code_1",

             "crime_code_2",

             "crime_code_3",

             "crime_code_4",

             "cross_street",

         ],

         axis=1,

         inplace=True,

     )

     

     # 删除存在缺失值的行

     data.dropna(inplace=True)

     ```



   - 沿用get_usefulData_feature_label中的fill_the_blank，不再使用month_day、specific_time，改用month、day、hour、minute



   - 所有LabelEncoder先不要做（留到[@Why-unable](https://github.com/Why-unable)做完数据可视化后再做）



   - 全都放在data中（后续尽量不要修改data，使用copy()，也不用写入.csv文件了）



3. Data Overview（还是[@ArronVague](https://github.com/ArronVague)做）



4. [@Why-unable](https://github.com/Why-unable)的算法，再扯皮一下结果



5. [@ArronVague](https://github.com/ArronVague)的算法及汇总



## 特征选择



month 月份（由date_occured拆分而来）



day 日期（由date_occured拆分而来）



hour（由date_occured拆分而来）



minute（由date_occured拆分而来）



area (area_name) 地区



victim_age 受害者年龄



victim_sex 受害者性别



victim_descent 受害者血统



latitude 纬度坐标



longitude 经度坐标



## 标签



crime_code (crime_descroption) 犯罪描述（训练效果差）



premise_code (premise_description) 遇害地点（如酒店、夜总会等）



weapon_code (weapon_description) 武器（训练效果较好）



status (status_descroption) 案件状态（训练效果好）



## 没用的特征



~~division_number 编号~~



~~date_reported 报告日期~~



~~reporting_district报告地点~~



~~part 犯罪事件的部分号~~



~~modus_operandi 作案手法~~



crime_code_1/2/3/4 犯罪编号



location 详细地址



cross_street 临近街道



## 算法性能汇总



|                     | crime_code | premise_code | weapon_code | status |

| ------------------- | ---------- | ------------ | ----------- | ------ |

| Logistic Regression | 0.28       | 0.27         | 0.54        | 0.61   |

| MLP                 | 0.28       | 0.27         | 0.54        | 0.61   |

| k-NN                | 0.22       | 0.22         | 0.47        | 0.51   |

| Naive Bayes         | 0.03       | 0.02         | 0.04        | 0.23   |

| Decision Tree       | 0.19       | 0.26         | 0.35        | 0.48   |

| Classifier Chain    | 0.19       | 0.26         | 0.35        | 0.48   |



## 参与贡献



1. clone仓库

2. 以main branch为基础new branch

3. 在新建分支上编写代码

4. commit代码到本地

5. publish branch到仓库

6. 创建pull request

7. 经审核人员审核后merge到main branch



## 贡献者



[![GitHub contributors](https://img.shields.io/github/contributors/ArronVague/Los-Angeles-Crime.svg)](https://github.com/ArronVague/Los-Angeles-Crime/graphs/contributors)



## 参考文献



[Los Angeles Crime Data Quick EDA 🦹🏼‍♂️](https://www.kaggle.com/code/guslovesmath/los-angeles-crime-data-quick-eda)



- 每个区域犯罪分布



- 犯罪状态统计



- 受害者血统



- 区域



[CrimeSolver Predictor](https://www.kaggle.com/code/safronov00/crimesolver-predictor#2.-Clean-Data)



- data overview

  - victim age

  - monthly crime（准备做成365天的，不看年份）

  - hour of the day

  - victim sex and descent distribution

  - top 10 crime descriptions and weapons used in crimes