Merge branch 'master' of github.com:dssg/dssg-landbank-project

Author: tplagge

README.md

@@ -6,14 +6,18 @@
 
 ####Data
 - Institute for Housing Studies (IHS) at DePaul University datasets
-- Cook County Assessor file
-- County shapefile
+- Cook County GIS and other agencies
+- County and City public data
+- Census
+- EPA
 
 ####Required Software
 - GIS software: ArcGIS 10.1 or QGIS
 - Postgres 9.1
 - PostGIS
 - Python 2.7.x
+- Numpy  1.7.1
+- Sympy  0.7.2
 
 ####Setup
 - Install Postgres and PostGIS

analysis/#prepForc.R#

@@ -6,3 +6,118 @@ neighb4comp(d=d)
 neighb4comp(d=d)
 quit()
 n
+d <- read.table(file="/home/evan/Documents/chicago/dssg-landbank-project/analysis/single_family_defaults_by_ca_over_time.csv",header = FALSE)
+d <- read.table(file="/home/evan/Documents/chicago/dssg-landbank-project/analysis/single_family_defaults_by_ca_over_time.csv",header = FALSE)
+d <- read.table(file="/home/evan/Documents/chicago/dssg-landbank-project/analysis/single_family_defaults_by_ca_over_time.csv",header = FALSE,sep = ",")
+head(d)
+sfneighb4comp(d=d)
+library(ggplot)
+library(ggplot2)
+sfneighb4comp(d=d)
+head(d)
+quit()
+n
+list.files()
+list.files(,pattern = pattern = "^[CA]", full.names = TRUE, ignore.case = TRUE)
+list.files(,pattern = "^[CA]", full.names = TRUE, ignore.case = TRUE)
+d <- read.table(file="/home/evan/Documents/chicago/dssg-landbank-project/analysis/CA_ptype_median_bldg_size.csv",header = TRUE,sep=",")
+d[1:10,]
+d[1:20,]
+d[1:30,]
+d[d['PTYPE2011']=='SINGLE FAMILY',]
+d1 <- d[d['PTYPE2011']=='SINGLE FAMILY',]
+community_area <- str_trim(d[,1],side = "both")
+community_area <- str_trim(d[,1],side = "both")
+d1<-data.frame(community_area,d1[,-1])
+community_area <- str_trim(d1[,1],side = "both")
+d1<-data.frame(community_area,d1[,-1])
+d1
+d1[1:100,]
+community_area==is.character
+str(community_area)
+d[446,]
+d[446,1] <- 'OHare'
+levels(d[,1])
+levels(d[,1])[60:63] <- c("OHare1","OHare2","OHare3","OHare4")
+community_area_labels<- str_trim(levels(d[,1]),side = "both")
+community_area_labels
+community_area_f <- factor(as.numeric(community_area),levels = 1:86,labels = community_area_labels)
+community_area_f
+as.numeric(community_area)
+community_area
+community_area[59]
+community_area[59] <- "OHare"
+d1
+nrow(d1)
+length(community_area)
+d1 <- data.frame(community_area,d1[,-1])
+head(d1)
+library(ggplot2)
+p <- ggplot(data=d1,aes(community_area,sqft_bldg_11))+geom_bar()
+print(p)
+p <- ggplot(data=d1,aes(community_area)+geom_bar()+stat_bin()
+)
+p <- ggplot(data=d1,aes(community_area))+geom_bar()+stat_bin()
+print(p)
+p <- ggplot(data=d1,aes(community_area,sqft_bldg_11,fill=sqft_bldg_11))+geom_bar()
+print(p)
+p <- ggplot(data=d1,aes(community_area,fill=sqft_bldg_11),stat="bin")+geom_bar()
+print(p)
+p <- ggplot(data=d1,aes(community_area,fill=sqft_bldg_11),stat="identity-")+geom_bar()
+print(p)
+p <- ggplot(data=d1,aes(community_area,fill=sqft_bldg_11),stat="identity")+geom_bar()
+print(p)
+p <- ggplot(data=d1,aes(community_area,sqft_bldg_11),stat="identity")+geom_bar()
+print(p)
+p <- ggplot(data=d1,aes(x=community_area,y=sqft_bldg_11),stat="identity")+geom_bar()
+print(p)
+p <- ggplot(data=d1,aes(x=community_area,y=sqft_bldg_11,fill=community_area),stat="identity")+geom_bar()
+print(p)
+d <- read.table(file="/home/evan/Documents/chicago/dssg-landbank-project/analysis/CA_ptype_median_bldg_psf.csv",header = TRUE,sep=",")
+head(d)
+d1 <- d[d['PTYPE2011']=='SINGLE FAMILY',]
+nrow(d1)
+nrow(d)
+nrow(d1)
+nrow(d2)
+nrow(d3)
+d1 <- d[d['PTYPE2011']=='SINGLE FAMILY',]
+colnames(d1)
+d2 <- data.frame(community_area,d1["sqft_bldg_11","bldg_assmt_11_psf","sqft_land_11","land_assmt_11_psf"])     
+head(d2)
+head(community_area)
+d3 <- d1["sqft_bldg_11","bldg_assmt_11_psf","sqft_land_11","land_assmt_11_psf"]
+d3 <- d1["sqft_bldg_11","bldg_assmt_11_psf","sqft_land_11"]
+d3 <- d1[,c("sqft_bldg_11","bldg_assmt_11_psf","sqft_land_11","land_assmt_11_psf")]
+head(d3)
+d3 <- data.frame(community_area,d1[,c("sqft_bldg_11","bldg_assmt_11_psf","sqft_land_11","land_assmt_11_psf")])
+head(d3)
+library(plyr)
+adply(d3,1,transform,singleFamSellPrice=((sqft_bldg_11*bldg_assmt_11_psf)+(sqft_land_11*land_assmt_11_psf)))
+p <- ggplot(data = d,aes(singleFamSellPrice))+geom_boxplot()
+print(p)
+colnames(d)
+p <- ggplot(data = d3,aes(singleFamSellPrice))+geom_boxplot()
+print(p)
+colnames(d3)
+d3 <- adply(d3,1,transform,singleFamSellPrice=((sqft_bldg_11*bldg_assmt_11_psf)+(sqft_land_11*land_assmt_11_psf)))
+p <- ggplot(data = d3,aes(singleFamSellPrice))+geom_boxplot()
+print(p)
+p <- ggplot(data = d3,aes(y=singleFamSellPrice))+geom_boxplot()
+print(p)
+p <- ggplot(data = d3,aes(x=community_area,y=singleFamSellPrice))+geom_boxplot()
+print(p)
+d3 <- data.frame(d3,"sigle_family")
+head(d3)
+d3[,1:(ncol(d3)-1)]
+d3 <- data.frame(d3,housing_type="single_family")
+head(d3)
+d3 <- d3[,c(1:6,8)]
+head(d3)
+p <- ggplot(data = d3,aes(x=housing_type,y=singleFamSellPrice,fill=singleFamSellPrice))+geom_boxplot()
+print(p)
+p <- ggplot(data = d3,aes(x=housing_type,y=singleFamSellPrice))+geom_boxplot()
+print(p)
+colnames(d)
+quit()
+n