8  Little Rock Crime Dataset

The dataset was collected from the Open Data source called little rock’s data hub. This dataset is 13 MB with 80,926 rows/observation and 14 different columns. The principal motivation behind using this dataset is to create a precise model to anticipate crimes. It can be downloaded from here.

The dataset could be cleaned as follows.

Data Cleaning and wraggling steps:

  1. Remove all the duplicates in the column INCIDENT_NUMBER.

  2. Set INCIDENT_NUMBER to be the index to uniquely identify offenses.

  3. Fill the missing values on WEAPON_TYPE to be NO WEAPON. Fill all other missing values by 0.

  4. Change 1 in WEAPON_TYPE to be UNKNOWN.

  5. Reassigned the columns to their respective types, e.g. numbers are numeric types and characters are strings.

  6. Parse the INCIDENT_DATE to find out the year, month, day, hour and min and create columns accordingly. Note that AM/PM is taken into consideration when computating hours.

  7. Create a column CRIME_TYPE based on the following rules:

    • If the OFFENSE_CODE contains 23, the crime is Non-Violent.
    • If the OFFENSE_DESCRIPTION contains THEFT, the crime is Non-Violent.
    • All other crimes are Violent Crime.

  8. Create a column RISK_TYPE based on the following rules:

    • If the crime is Non-Violent and NO WEAPON, it is Low Risk.
    • All other crimes are High Risk.