Project Journal, Week 9
Hello everyone! I am glad to deliver this week’s blog to you guys. This blog will be a sequel to our Chicago vs. Temperature blog post. If you haven’t check that out, please follow this link:
Overall Analysis
We made a mistake in our previous analysis. We thought about the unusual similar pattern across different crime types and time periods but the visualization of the number of incidents over time suggests a strong pattern so we ignored the issue.
However, we did sum up all the incidents with the same average temperature and we did not account for the fact that there will be more 75 degree days than -20 degree days. Therefore, we visualized the net number of incidents instead of the average number of incidents for each temperature. Figure 2 below is the correct visualization with Average Incidents as the vertical axis and the Average Temperature as the horizontal axis.
Figure 2 shows a very nice correlation with the r-squared score of 0.67. The figure suggests that the higher the temperature the more likely that crimes can happen. The average number of crimes gets a bit lower when it reaches very high temperatures, but it is not as low as those of very low temperatures.
Different Time Periods
Let’s take a look at different slices of the data. Figure 3 shows the average incidents vs. average temperature in different periods of time.
If we look back at Figure 1, we can see a downhill trend in crime rates in recent years. This trend reflects clearly in Figure 3, as the height of the graph gets lower as it gets further into the future. We also see a similar correlation between crime rates and the average temperature in this figure.
Different Types of Crime
Additionally, we looked into the pattern of individual types of crimes and visualized this pattern. Apart from those without sufficient data, all types of crime share the same temperature pattern that we have seen above. This is interesting because we expected that some crime types would be committed more in extreme weather conditions. One of the hypothesized crime types would be Motor Vehicle Theft because we assume that criminals would pick times when people are not around to avoid being detected.
Yet, Figure 4 suggests that Motor Vehicle Theft incidents follow the pattern of other incident types. It seems like Motor Vehicle Theft is less dependent on the temperature because the slope of the regression line is closer to 0.
That is what we have for this week’s blog. Make sure to hit us up or comment down below if you have some thoughts to share!
