Wednesday, October 28, 2015

Lab 3 - Significance Testing


1.


2. Ground Nuts
Probability: 35.24
z Value: -.9346

Fail to reject null hypothesis because it lies in the range of normal distribution.

Cassava
Probability: 0.0348
z Value: -2.1127

Reject the null hypothesis. -2.1127 falls too far away from the normal distribution of -1.96

Beans
Probability: 0.324
z Value: 2.1429

Reject null hypothesis. The samples are too different.

There are a few similarities amongst the data including 2/3 data sets fall outside of the significant range. As far as differences go each data set had a completely different Z-Score. This shows that among the three different crops there are some very different amounts.

3. Null Hypothesis: With a 95% confidence interval it shows there is not a difference between the average number of people per party comparing 1960 and 1985.

Probability:
1960: 2.8
1985: 3.7 (Sample size: 25)

T Test
(3.7 – 2.8) / (1.45/sqrt 25) = (0.9)/(0.29) = 3.1034
df=25-1=24

1.711

There is a significant difference between the 1985 sample and the 1960 party size meaning the null hypothesis is rejected.


Wednesday, October 7, 2015

Assignment 2

Introduction:

     Local residents of Eau Claire have begun to complain about the amount of disruption college age residents have been causing especially in the Water Street Area. There are a lot of factors that may contribute to the issue but most believe it could be related to the amount of alcohol consumed in the area.
     Using crime data provided by the Eau Claire Police Department once can start making correlations with the proximity of the arrests to certain block groups and even more so their proximity to establishments that serve alcohol. Having multiple years of data is important because it can be used to see if patterns can be associated with this behavior. In the data there aren't specifics as to why the crimes occurred but using spatial tools they can be analyzed nonetheless.

Methods:

     To analyze the data I used a variety of spatial tools to dig deeper in to the data I was presented with. Because this data all revolves around the location of crimes committed it was important that the distribution of crime locations be looked it. The first facet I looked in to was the mean centers of the crimes. (Figure 1) (Figure 2) Mean centers are important because it takes all the points inputted and finds the mean location or geometric center. Weighted mean centers are also important because they are similar to mean centers but unlike mean centers it takes in to account the number of instances at one certain point as well.
Figure 1 - Mean Centers for Disorderly Conducts in Eau Claire - 2003


Figure 2 - Mean Centers for Disorderly Conducts in Eau Claire - 2009

     The data above is from two different years although if you quickly glance at them there are some glaring similarities, especially with the various mean centers. To compare the two years I also created a map that includes all the data from the above two maps. (Figure 3) The mean centers show the average location of where all the crimes were committed.


Figure 3 - Combined Data for 2003 & 2009

     The next tool I used to analyze the data was standard distance. Standard distance is perfect for the data we have because it shows us the area with the highest occurrences.  The standard distance ring covers 68% of the points so more than half the points can be found inside of the ring. Standard distance is a tool used to show the highest concentration of points, the results can be found below in (Figure 4) (Figure 5).

Figure 4 - Weighted Standard Distance around the Weighed Mean Center - 2003

Figure 5 - Weighted Standard Distance around the Weighted Mean Center - 2009
  
     Once again the two maps are very similar comparing 2003 to 2009. Here is a map comparing the two (Figure 6).

Figure 6 -  Combined results of Figure 4 and 5.
 
     The standard distances are overlapping almost entirely so between the two years of data there is almost no change in weighted means.

     Many of the residents believe the occurrences of disorderly conducts can be related to the number of bars in an area as there are often high concentrations of college students. To show this information I used the block data for Eau Claire and included the bar locations within each individual block. For comparison I also included the standard deviation of disorderly conducts within each block (Figure 7). This data was from 2009. Standard deviations are good indicators of how much variation there was from block to block and can highlight where the high volume of arrests occurred.
Figure 7 - Standard deviation representation of arrests per block group with bar locations - 2009.
Results:

     After analyzing the data there are some interesting trends that jump out. As (Figure 1) and (Figure 2) show you can see that the weighted means pull away slightly southwest indicating that there are higher occurrences at specific addresses in that direction. These points are also closer to Water St. If you compare 2003 to 2009 using (Figure 3) you may notice that although similar both the mean centers and weighted mean centers are trending slightly more north in 2009 compared to 2003. In order to see if this trend has been occurring since 2003 we would need more data.
     Upon examination of the standard distances in (Figures 3-6) we can see that the size of ellipse in 2009 is smaller than 2003 meaning there is more of a concentration in arrests surrounding the mean.
     (Figure 7) is very telling of the current situation in Eau Claire. The block group containing Water St. has as standard deviation of over 2.5 meaning it has a high variation compared to the other block groups. Coincidentally there is a higher amount of bars in this block as well along with a high student population although that does not necessarily guarantee the two are related.

I was also tasked with finding the Z scores of three certain block groups.

The first block group was 41, in this block group notable structures are Oakwood Mall and some residential neighborhoods.
Disorderly Conducts: 10
Mean: 5.3593
Standard Deviation: 7.8149

Z-Score = (10 – 5.3593)/7.8149
Z-Score = 0.5938
 
The second block group was 46, which could be considered "Old Downtown."
Disorderly Conducts: 40
Mean: 5.3593
Standard Deviation: 7.8149
 
Z-Score = (40-5.3593)/7.8149
Z-Score = 4.4326
 
The last block group was 57 which can be found north west of block 46.
Disorderly Conducts: 1
Mean: 5.3593
Standard Deviation: 7.8149
 
Z-Score = (1-5.3593)/7.8149
Z-Score = -0.5578


 
Additionally, I was asked to figure out some probabilities for the Eau Claire block groups. With the current trends, we can figure out the probability of a certain amount of disorderly conducts to occur in a block group.
Z-Score = -0.52
-0.52 = (X – 5.3593) / 7.8149
-4.0637 = X – 5.3593
X = 1.2956
This tells us that 70% of the time the amount of disorderly conducts per block group will exceed 1.2956. 
 
On the opposite side of the spectrum:
Z-Score = 0.84
0.84 = (X – 5.3593) / 7.8149
6.5645 = X – 5.3593
X = 11.9238
Meaning 20% of the time the amount of disorderly conducts per block group will exceed 11.9238. 
 
Conclusion: 
     By looking at the results you can see that Water St. has a large impact on the results as the mean centers pull towards it. The standard distance ellipses also contain much of which is student housing around Water St. near the campus. It is reasonable for the complaining citizens to be argue that college students are the primary cause of the "ruckus." Looking at the standard deviation for that block group you can see that it is over 2.5 standard deviations away from the mean.
 
     However, Water St. was not the only area with a significant amount of disorderly conducts. Many were filed near the police station on Lake St. It is hard to tell if the actual crime was committed there or if they were just brought to the station to be processed. There was also a significant amount of disorderly conducts near the confluence, an area that also contains a large amount of bars looking at (Figure 7).  
 
     It is easy to make assumptions just by looking at the results but unfortunately disorderly conducts is a very broad and varying charge. You can have anything from public drunkenness too as simple as loitering in a certain area. Not all disorderly conducts have to involve alcohol assumption.  

 
     There is no simple fix to this issue, higher police representation may help but the biggest factors are educating students and having those student be personally accountable for themselves and their friends. I would take in to consideration the specific location of where the complaints are coming from and possibly up police representation in that areas in hopes that it lowers the disturbances. As you can see from the maps there are many areas in Eau Claire that do not experience disorderly conducts at all so the majority of police resources should be focused on the problem blocks in hopes of reducing crime and disturbances.
 
     Based on the information provided I would say yes the college students are a large factor in the in the disturbance the citizens are complaining about. However, I would be very cautious for blaming alcohol consumption as there is no conclusive evidence of that, just trends that may or may not be coincidental. No one will argue that fact that a lot of disorderly conducts do happen near Water St. and in an area that houses a lot of students though.