Reference growth with time. Many Linear Regressions with Python. Create R plots in a loop; layout with grid.arrange

fit.py

#!/usr/bin/env python

import sys
import datetime
import re
from pprint import pprint
import math
from scipy import stats
# cat nikeplot.csv | ./fit.py > fit.csv

# Dates
date1RE = re.compile("[0-9]{4}-[0-9]{2}-[0-9]{2}.[0-9]{2}:[0-9]{2}:[0-9]{2}")
date1Parse = "%Y-%m-%d %H:%M:%s"

# Delimiters
inDelim = ","

# read input
rowCount = 0
x = []
y = []
last=''
print ",".join(["url","r","m","b","alpha","velocity"])
for r in sys.stdin:
	rowCount += 1
	row = r.strip().split(inDelim)
	if last <> row[0]:
		#print x,y
		if len(x) > 4:
			m,b,r,p,stdr = stats.linregress(x,y)
			alpha = m
			velocity = math.exp(b/m)
			print ",".join([str(row[0]), str(r), str(m), str(b), str(alpha), str(velocity)])
		x = []
		y = []
		last = row[0]
	try:
		x.append(math.log(float(row[1])))
		y.append(math.log(float(row[2])))
	except ValueError:
		pass

nikeplot.csv

plot.r

#!/usr/bin/env Rscript

library(ggplot2)
library(gridExtra)

##################################
#read data file with headers
Y <-read.delim("./fit.csv", sep=",", header=TRUE)

plot(Y)
summary(Y)

#######
p1 <-ggplot( data=Y) +
        geom_histogram(aes(alpha), binwidth=0.025) +
        xlab("alpha") + ylab("# of URLs")
p2 <-ggplot( data=Y) +
        geom_histogram(aes(velocity), binwidth=0.5) +
	scale_x_log10() + 
        xlab("velocity") + ylab("# of URLs")

png(filename = "./fit_hist.png", width = 500, height = 700, units = 'px')
 print(
   grid.arrange(p1, p2, ncol = 1, main="NikeFuckYeah - Log-log Fit")
 )
dev.off()

p3 <-ggplot( data=Y ) +
        geom_point(aes(velocity, alpha), color="red" ) +
        xlab("velocity") + ylab("alpha (fan-out)") + 
	scale_x_log10() + 
	scale_y_log10() +
	stat_smooth()

png(filename = "./cor.png", width = 500, height = 500, units = 'px')
 print( p3 )
dev.off()


topFan <- Y[order(Y$alpha, decreasing=T),]
topSpeed <- Y[order(Y$velocity, decreasing=T),]

tf <- topFan[5:20,1]
ts <- topSpeed[5:20,1]

#print(tf)
#print(ts)

##################################
#read data file with headers
X <-read.delim("./nikeplot.csv", sep=",", header=TRUE)

plot(X)
summary(X)

####### For loop create a list of plots that are then placed on a grid
plot.list <- list()
for (urli in tf) {
 t <- X[X$url==urli & X$time > 0, ]
 t$ltime <- log10(t$time)
 t$lcount <- log10(t$count)
 p <- ggplot(data=t, aes(ltime,lcount)) + stat_smooth(method="lm") + geom_line() 
 plot.list[[length(plot.list) + 1]] <- p
 print( p)
}
arg.list <- c(plot.list, 4, 4)
names(arg.list) <- c( "a","b","c","d",  "e","f","g","h",  "i","j","k","l",  "m","n","o","p"  ,"nrow","ncol") 

png(filename = "./top_fits.png", width = 600, height = 800, units = 'px')
  print(
    do.call(grid.arrange, arg.list)
  )
dev.off()
#######

p <-ggplot(aes(time, count), data=X) +
        geom_point(aes(color=url)) +
	coord_flip()+
	scale_y_log10() + 
	scale_x_log10() + 
	xlab("Time (sec from 0)") + ylab("Note Count")
ggsave(file="./nikefuckyeah_geom.png", width=9, height=12, dpi=100)

p <-ggplot(aes(time, count), data=X) +
        geom_point(aes(color=url)) +
	coord_flip()+
	scale_y_log10() + 
	xlab("Time (sec from 0)") + ylab("Note Count")
ggsave(file="./nikefuckyeah_exp.png", width=9, height=12, dpi=100)

p <-ggplot(aes(time, count), data=X) +
        geom_point(aes(color=url)) +
	coord_flip()+
	scale_x_log10() + 
	xlab("Time (sec from 0)") + ylab("Note Count")
ggsave(file="./nikefuckyeah_weird.png", width=9, height=12, dpi=100)