Crime in US
Descriptive Analytics
Nicolas Calderon
Advanced Analytics
Agenda
- Access the Problem Specification
- Creating a Latent Variable
- Analysis
- Fitting a Model: PCR-FAR vs PLSR
Access the Problem Specification
The New Latent Variable Criminality Level
Creating a new Criminality Level
Creating a new Criminality Level
\[
\small
\begin{equation*}
\begin{aligned}
Lantent_{i} & =-0.362 * Murders_{i} - 0.336 * Rapes_{i} - 0.407 * Robberies_{i}\\
& - 0.365 * Assults_{i} - 0.392 * Burgleries_{i} - 0.31 * larcenies_{i}\\
& - 0.351 * Authothefts_{i} - 0.286 * Arson_{i}
\end{aligned}
\end{equation*}
\]
Crime Level in Scale 0-1
\[
\small
CrimeLevel_{i} = \frac{Lantent_{i}-min(Lantent_{i})}{max(Lantent_{i})-min(Lantent_{i})}
\]
| Crime Categories | # Communities | Percentile |
|---|---|---|
| Low | 554 | 0-25 |
| Medium | 1107 | 25-75 |
| High | 554 | >75 |
Spatial View of Criminality
|
|
|
Bi-Variate Analysis
Economical Achievement
Economical Achievement
Economical Achievement
Population Composition
Social Behavior
Fitting a Model:PCR-FAR Vs PLSR
Models Estructure
