This document evaluates wheter we predict the proportion of within state shipments with NASS covariates? The following county-level NASS covariates are considered for prediction: the inventory of cattle, including claves; the number of operations with inventory; the net income per operation, and the number of operations receiving water. All covariates were downloaded from <> from the 2012 NASS census. They were selected for consideration based on two criteria. First, these data have few counties with censored information. Second, they represent information on industry infastructure, econmics, and environmental conditions.

Below, I first plot all covariates considered and report statistical analyses to maximize prediction.

I. NASS covariates considered:

Each covariate is plotted by percentiles, where colors scales represent 10% increases in the rage of the data. Color scales are consistent between the three states considered (California, Iowa, Montana, and Wyoming)

Inventory of cattle, including calves

Operations with inventory of cattle, including calves

Inventory of Cattle, Cows, Beef

Operations with inventory of Cattle, Cows, Beef

Inventory of Cattle, Cows, Milk

Operations with inventory of Cattle, Cows, Milk

Income, net cash from farm, in dollars per operation

Operations receiving water

II. Analyses predicting the proportion of within state shipments.

Below, I show scatterplots for each covariate individually.

Operations with inventory

Inventory of cattle

Income, net cash farm measured in dollars per operation

Operations receiving water

Model dependent variable

df2 <- df[c(-19, -21),]
test.mod <- glm(outships ~ cattleop + cattleinv + money + water, offset = log(df2$totalships), data = df2, family = quasipoisson(link = "log"))
summary(test.mod)
## 
## Call:
## glm(formula = outships ~ cattleop + cattleinv + money + water, 
##     family = quasipoisson(link = "log"), data = df2, offset = log(df2$totalships))
## 
## Deviance Residuals: 
##      Min        1Q    Median        3Q       Max  
## -27.4714   -4.8831   -0.2031    2.7992   22.9861  
## 
## Coefficients:
##              Estimate Std. Error t value Pr(>|t|)    
## (Intercept) -0.881261   0.038004 -23.188   <2e-16 ***
## cattleop    -0.024490   0.034823  -0.703    0.483    
## cattleinv    0.008167   0.017688   0.462    0.645    
## money        0.043919   0.025468   1.725    0.087 .  
## water        0.019854   0.043667   0.455    0.650    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for quasipoisson family taken to be 65.82874)
## 
##     Null deviance: 9071.1  on 132  degrees of freedom
## Residual deviance: 8663.7  on 128  degrees of freedom
##   (10 observations deleted due to missingness)
## AIC: NA
## 
## Number of Fisher Scoring iterations: 4