import pandas as pd
import numpy as np
import pandas.rpy.common as com

galton = com.load_data('galton', package='UsingR')
galton.head()

f, (ax1, ax2) = subplots(ncols=2)

ax1.hist(galton['child'], bins=100)
ax2.hist(galton['parent'], bins=100)

f.tight_layout();

f, ax = subplots(ncols=1)

ax.hist(galton['child'], bins=100)

meanChild = galton['child'].mean()

ax.plot(np.repeat(meanChild, 100), np.linspace(0, 150, num=100), 'r', linewidth=5);

plot(galton['parent'], galton['child'], 'ob');

def jitter(series, factor):
    z = float(series.max()) - float(series.min())
    a = float(factor) * z / 50.
    return series.apply(lambda x: x + np.random.uniform(-a, a))
   
np.random.seed(1234)
plot(jitter(galton['parent'], factor=1), jitter(galton['child'], factor=1), 'ob', alpha=.5);

from statsmodels.formula.api import ols

lm = ols('child ~ parent', galton).fit()

f, (ax1, ax2) = subplots(ncols=2)

# plot fitted values
ax1.plot(galton['parent'], galton['child'], 'ob')
ax1.plot(galton['parent'], lm.fittedvalues, 'r', linewidth=3);

# plot residuals
ax2.plot(galton['parent'], lm.resid, 'ob')
ax2.plot(galton['parent'], np.repeat(0, len(galton['parent'])), 'r', linewidth=3)

f.tight_layout();