R Week 4 Solution

Data : US States Production - A panel of 48 observations from 1970 to 1986

References, source, and metadata http://vincentarelbundock.github.io/Rdatasets/doc/plm/Produc.html

The following R code and graphs is an exploration into relationships between public and private spending by state and unemployment rates between the years of 1970 and 1986. This exploration concludes with an analysis of a subset of states with the smallest and greatest range of private versus public spending during this time and an attempt at finding a relationship between the data.

1. We load the data and create a private versus public spending ratio for each year for each state.

df<-read.csv("http://vincentarelbundock.github.io/Rdatasets/csv/plm/Produc.csv", header= TRUE, sep=",")
options(warn=-1)
df$priv_over_pub <- with(df, pc / gsp)
df_sub<-subset(df, select=c(state, year, unemp,priv_over_pub))

2. We plot the data into a histogram to see any initial trends between the currently reduced data.

3. A boxplot of the data per year reveals some further insight into the trends and behavior of spending per state over this time period.

4. In order to reduce the data set, we decide to calculate the ranges between the lowest and highest ratio values per state for each year.

min_vals <-aggregate(priv_over_pub ~ state, df_sub, function(x) min(x))
max_vals <-aggregate(priv_over_pub ~ state, df_sub, function(x) max(x))
names(min_vals)[1:2]<-c("state","min")
names(max_vals)[1:2]<-c("state","max")
maxmin=merge(min_vals,max_vals,by='state')
maxmin$range <- (maxmin$max-maxmin$min)
maxmin[order(maxmin$range),]

##             state       min       max     range
## 47      WISCONSIN 0.8408928 0.9446676 0.1037748
## 18       MARYLAND 0.6775637 0.7957770 0.1182133
## 44       VIRGINIA 0.7045575 0.8358151 0.1312577
## 28     NEW_JERSEY 0.6665610 0.7979046 0.1313436
## 19  MASSACHUSETTS 0.5885171 0.7257024 0.1371852
## 31 NORTH_CAROLINA 0.8600003 0.9991763 0.1391760
## 22    MISSISSIPPI 1.1753641 1.3199585 0.1445944
## 6     CONNECTICUT 0.6147361 0.7601229 0.1453868
## 4      CALIFORNIA 0.6501949 0.7994933 0.1492984
## 42           UTAH 0.9722568 1.1240468 0.1517900
## 9         GEORGIA 0.8863980 1.0386686 0.1522707
## 40       TENNESSE 1.0015547 1.1592838 0.1577291
## 23       MISSOURI 0.8731555 1.0320287 0.1588732
## 30       NEW_YORK 0.5628860 0.7220925 0.1592065
## 36   PENNSYLVANIA 0.8135829 0.9759466 0.1623637
## 33           OHIO 0.8608755 1.0238819 0.1630064
## 37   RHODE_ISLAND 0.5534691 0.7208513 0.1673822
## 29     NEW_MEXICO 1.3293738 1.5013790 0.1720052
## 41          TEXAS 1.2222095 1.3958253 0.1736158
## 45     WASHINGTON 0.9124592 1.0869291 0.1744699
## 10          IDAHO 1.1662234 1.3439705 0.1777470
## 5        COLORADO 0.8448481 1.0388615 0.1940134
## 38 SOUTH_CAROLINA 1.1326606 1.3303239 0.1976633
## 11       ILLINOIS 0.7878404 0.9857199 0.1978795
## 3        ARKANSAS 1.1502799 1.3642801 0.2140002
## 8         FLORIDA 0.7308522 0.9586563 0.2278041
## 12        INDIANA 1.0549468 1.2860535 0.2311067
## 1         ALABAMA 1.1990431 1.4387445 0.2397015
## 15       KENTUCKY 0.9124670 1.1652944 0.2528274
## 21      MINNESOTA 0.8885572 1.1454508 0.2568937
## 17          MAINE 0.9002072 1.1605149 0.2603077
## 35         OREGON 0.9380360 1.1988304 0.2607944
## 14         KANSAS 1.2269931 1.4891513 0.2621582
## 24        MONTANA 1.6908443 1.9567083 0.2658640
## 43        VERMONT 0.8403514 1.1081802 0.2678288
## 25       NEBRASKA 1.1629348 1.4314748 0.2685400
## 13           IOWA 1.0688918 1.3560258 0.2871340
## 20       MICHIGAN 0.7667430 1.0670583 0.3003153
## 7        DELAWARE 0.8010332 1.1058352 0.3048020
## 2         ARIZONA 0.9053479 1.2228323 0.3174845
## 39   SOUTH_DAKOTA 1.2905422 1.6113060 0.3207637
## 34       OKLAHOMA 1.1509080 1.4805259 0.3296179
## 27  NEW_HAMPSHIRE 0.6471302 0.9810807 0.3339505
## 46  WEST_VIRGINIA 1.3458595 1.7257098 0.3798503
## 26         NEVADA 1.1625266 1.5481512 0.3856246
## 16      LOUISIANA 1.4756784 1.9650905 0.4894121
## 32   NORTH_DAKOTA 1.5610037 2.2324990 0.6714953
## 48        WYOMING 1.7308232 2.4940672 0.7632439

5. We can then subset the two dataframes into the 5 greatest, and 5 lowest ranges per state per ratio.

maxmin2 <- subset(maxmin, range > 0.35 | range < 0.138,select=c(state, range))
require(plyr)

## Loading required package: plyr

combinedData <- join(maxmin2, df, by='state', type='left', match='all')
minstate <- subset(combinedData, range < 0.138,select=c(state, year, range, unemp, priv_over_pub))
maxstate <- subset(combinedData, range > 0.35 ,select=c(state, year, range, unemp, priv_over_pub))

6. A plot of our lowest or most “consistent” ranges of subset [Wisconsin, Maryland, Virginia, New Jersey, Massachusetts] shows a range in unemployment.

7. A plot of our lowest or most “inconsistent” ranges of subset [West Virginia, Nevada, Louisiana, North Dakota, Wyoming] shows a range in unemployment and trend that differs from the earlier graph.

8. We cannot conclude much from these findings, but we may be able to improve our initial hypothesis and parameters. We can see a relationship between and erratic public vs private state budget and unemployment issues during this time. These relationships are certainly not directly related, but would be part of a much bigger scenario.