python snippets.py

# push subfolder in git repo called "app" to heroku
git push heroku `git subtree split --prefix app master`:master --force

###

pd.set_option('max_colwidth', 400)

from IPython.display import display
with pd.option_context("display.max_columns", 50):
    display(df.head(2))
    
from IPython.core.debugger import Tracer; Tracer()() 

###

from __future__ import division
import pandas as pd
import warnings
import seaborn as sns
import matplotlib.pyplot as plt
from pylab import rcParams
%matplotlib inline

warnings.filterwarnings("ignore", category=DeprecationWarning)
sns.set_style("whitegrid")
sns.set_context("poster")
rcParams['figure.figsize'] = 20, 5

from scipy.sparse import csc_matrix
from zipcode_mapping import zipcode_mapping

##

  # grep recursively through files of a certain type, such as .py files 
grep -R --include="*.py" "pattern" /path/to/dir

find . -name *simple*


##

def dummify(df, column):
    # from Darren's linear regression slides
    print '{} is your baseline'.format(sorted(df[column].unique())[-1])
    dummy = pd.get_dummies(df[column]).rename(columns=lambda x: column+'_'+str(x)).iloc[:,0:len(df[column].unique())-1]
    df = df.drop(column,axis=1) #Why not inplace? because if we do inplace, it will affect the df directly
    return pd.concat([df,dummy],axis=1)

##
with pd.option_context("display.max_rows", 200):
    display(df)