earino/data_science_hw1

data science, homework 1

Assingment:

For homework, please try creating a regression model using any dataset of your choice and tell us why you picked that model (e.g. absolute t values < 1, all significant relationships, etc..) along with sending the R script/commands you used to generate the model. Please use an iterative process. If you want to save your code from your console session, you can use the command "savehistory(file.choose())". Please send r script + brief explanation to Jason by Thursday of next week.

My Info:

I decided to create a merged data set taking the 2012 LA crime information data available at http://www.lasdhq.org/sites/CAASS/2012-PART_I_AND_II_CRIMES.csv

and the 1999 Average Household income of LA county available at http://www.laalmanac.com/LA/la09_zip.htm.

Using a number of perl scripts (i could have been neater, but I was just playing around) I took the crime information, grabbed the zip code, and simply incremented a counter for every entry in the data file. I then merged that data with the LA income data to create a new data file called merged_incomes_crimes.csv which was

zip, crimes, income

I then created a linear model in which I verified that, not shockingly, there is a negative correlation between household income in a zipcode and crimes commited in that zip code. Richer zip codes have fewer reported crimes.

Call:
lm(formula = income ~ ., data = merged_incomes_crimes)

Residuals:
   Min     1Q Median     3Q    Max 
-33078 -12502  -3395   8191  70894 

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept) 41746.865   1916.589  21.782   <2e-16 ***
crimes         -4.034      2.276  -1.772   0.0793 .  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 18750 on 103 degrees of freedom
Multiple R-squared:  0.02959,Adjusted R-squared:  0.02017 
F-statistic: 3.141 on 1 and 103 DF,  p-value: 0.0793