Skip to content

Latest commit

 

History

History
129 lines (79 loc) · 4.24 KB

README.md

File metadata and controls

129 lines (79 loc) · 4.24 KB

🦈 Global Shark Attacks data

Before modeling, data understand is the key part. Thus, we are going to do Exploratory Data Analysis (EDA) and trying to understand data fully. That is the main scope of the work. Let's follow the steps.

For fun, the global shark attacks dataset has been considered in this work.

☾ Table of contents

  • Data acquisition
  • Dataset
  • Preprocessing
  • Data mining : EDA, correaltion matrix, visualization
  • Features engineering

☺︎ Data acquisition

download data set from the kaggle link below

🦆 Kaggle data Global shark attacks file

☺︎ Dataset

First, you read csv file and check all variables.

df0 = pd.read_csv("attacks.csv",encoding = "ISO-8859-1")

#..... describe dataset 
print('================================================')
print(df0.describe)
print('================================================')
print('total size :',df0.shape)

☻ Variables

Total size of data is 25723 rows and 24 columns.

Case Number Date Year Type country Area Location Activity Name Sex Age Injury Fatal (Y/N)
2018.05.26.a 26-May-2018 2018 Unproviked USA Florida Daytona Standing male M 12 Minor injury to foot N
Time Species Investigator or Source pdf href formula href Case Number.1 Case Number.2 original order Unnamed: 22 Unnamed: 23
14h00 nan K. McMurray, Tracking Sharks.com 2018.05.26.a-DaytonaBeach.pdf http://sharkattackfile.net/spreadsheets/pdf_directory/2018.05.26.a-DaytonaBeach.pdf http://sharkattackfile.net/spreadsheets/pdf_directory/2018.05.26.a-DaytonaBeach.pdf 2018.05.26.a 2018.05.26.a 6294 nan nan

In details, it is necessary to understand all columns, contents, and values whether numerical or categorical.

  • Case Number : Date
  • Date : LITR, date when incident happended.
  • Year : LITR, year
  • Type : provoke, unprovoked, ...
  • Country
  • Area : area, espeically states in the USA
  • Location : exact location
  • Activity : what kinda activities when they were attacked by sharks
  • [delete] Name : name of victims
  • Sex : LITR, sex. Male or Female
  • Age : LITR, age
  • Injury : wounded parts in the attack
  • Fatal (Y/N) : seriousness
  • Time : Exact time or a time slot
  • Species : what kinda sharks
  • [delete] Investigator or Source : name who invested the incident
  • [delete] pdf : title of the investing file
  • [delete] href formula : directory of file
  • [delete, duplicate] href : same w/ href formula
  • Case Number.1 : Casenumber, same w/ date
  • [delete, duplicate] Case Number.2 : same w/ Case Number.1
  • [delete, duplicate]original order : order of incidents
  • [delete] Unnamed: 22 : all nan
  • [delete] Unnamed: 23 : all nan

☻ Issues

There are several erorrs including missing values, duplication, noise, categorical values, mixing values etc. Thus, data cleaning and preprocessing are strongly required and then different approaches should be considered for data analysis.

☺︎ Preprocessing

Before moving to EDA, preprocessing is required due to several issues

☻ Data cleaning

☻ Encoding

categorical data -> numerical data

☺︎ Data mining

☻ Data visualization

☻ Features correlation

☻ Reference

☻ image sources

  1. Giphy