def query_random(df):
"""Return a (sub-dataframe, namekey, placename) corresponding to a
particular namefamily. Also return a sample placename from that
namefamily"""
df = df.copy()
namekeys = name_rules.keys()
namekey = sample(namekeys, 1)[0]
subdf = get_fam(df, namekey)[0]
placename = subdf['name'].sample().values[0]
return subdf, namekey, placename
import pandas as pd; import numpy as np; from pandas import Series, DataFrame
get_ipython().magic(u'ls ')
get_ipython().magic(u'cd projects/')
get_ipython().magic(u'cd data-petting-zoo/')
get_ipython().magic(u'ls ')
from rules import name_rules, geofeat_rules
gbf = 'data/pristine/GB.txt'
gb = setgb(gbf)
other = setfam(gb)
from engine import setgb, setfam
colors = cycle(['orange', '#eeeeee', 'b', 'y', 'cyan', 'magenta', 'yellow', 'red', '#cccccc', 'green', 'pink'])
from itertools import cycle
colors = cycle(['orange', '#eeeeee', 'b', 'y', 'cyan', 'magenta', 'yellow', 'red', '#cccccc', 'green', 'pink'])
import matplotlib.pyplot as plt
for key in name_rules.keys():
data = getfamdf(gb2, key)
plt.scatter(data.long, data.lat, label=key, c=next(colors), s=80, alpha=0.4)
plt.legend(loc='lower right')
from engine import getfamdf
get_ipython().magic(u'matplotlib ')
for key in name_rules.keys():
data = getfamdf(gb2, key)
plt.scatter(data.long, data.lat, label=key, c=next(colors), s=80, alpha=0.4)
plt.legend(loc='lower right')
for key in name_rules.keys():
data = getfamdf(other, key)
plt.scatter(data.long, data.lat, label=key, c=next(colors), s=80, alpha=0.4)
plt.legend(loc='lower right')
gb = setgb(gbf)
