import pandas as pd
%pylab inline

def dateparser(datestring):
    return datetime.datetime.strptime(datestring,'%Y%m%d%H')

total= pd.read_csv('total-volume.csv',parse_dates=[0],date_parser=dateparser)
total.columns=["date","tweets"]

total.head()

total.index = total.date

pylab.rcParams['figure.figsize'] = (16.0, 8.0)
total.plot(marker="o",markerfacecolor="red")

perday = total.resample("W-Mon",how="sum").plot(kind="bar")
title("Tweets per Week")
xlabel("Sum of Tweets per Week")

model = pd.ols(y=log(total.tweets[:400]), x=total.tweets[:400].shift(-1), intercept=True)

print model.summary


fig=figure()
sub=fig.add_subplot(111)
plot(model.sm_ols.model.fit().fittedvalues,linewidth=0,marker="o")
plot(log(total.tweets[:400]),color="red",marker="o",linewidth=0)

t=total.tweets.tolist()
t1=total.tweets.shift(-1).tolist()


%load_ext rmagic
tweettoday = total.tweets.tolist()
tweetyesterday =  total.tweets.shift(1).tolist()

%%R -i tweettoday,tweetyesterday
require(ggplot2) 
print(qplot(x=tweettoday,y=tweetyesterday,alpha=.5))
print(summary(lm(tweettoday~tweetyesterday)))