In [1]:

    

%matplotlib inline 

import numpy as np
import pandas as pd

import matplotlib.pyplot as plt
import seaborn as sns

sns.set_context("poster")
sns.set_style("ticks")

from sklearn.feature_extraction.text import CountVectorizer, TfidfTransformer, TfidfVectorizer
from sklearn.linear_model import LinearRegression
from sklearn.cross_validation import cross_val_score
from sklearn_pandas import DataFrameMapper, cross_val_score


    

In [2]:

    

df = pd.read_csv('https://query.data.world/s/d3gals7qk5sz9l2o50oirszif')
df = df.rename(columns={
        "Revenues (M)": "Revenues"
    })
df = df.assign(Revenues=df["Revenues"].apply(lambda x: float(x[1:].replace(",", ""))))
df["Mission"] = df["Mission"].fillna("")


    

In [3]:

    

df.head()


    

    
Out[3]:






  

    

      

      
Name
      
Rank
      
Revenues
      
Mission
      
Mission Link
    
  
  

    

      
0
      
Walmart
      
1
      
482130.0
      
We save people money, so they can live better
      
http://corporate.walmart.com/
    
    

      
1
      
Exxon Mobil
      
2
      
246204.0
      
Exxon Mobil Corporation is committed to being ...
      
http://corporate.exxonmobil.com/en/company/abo...
    
    

      
2
      
Apple
      
3
      
233715.0
      
Apple designs Macs, the best personal computer...
      
http://www.inc.com/jim-schleckser/apple-s-bori...
    
    

      
3
      
Berkshire Hathaway
      
4
      
210821.0
      

      
NaN
    
    

      
4
      
McKesson
      
5
      
181241.0
      
Together with our customers and partners, we a...
      
http://www.mckesson.com/about-mckesson/who-we-...
    
  

In [4]:

    

np.log10(df.Revenues).hist(bins=100, log=True)
plt.xlabel("$log_{10}(Revenue)$")


    

    
Out[4]:





<matplotlib.text.Text at 0x7fe00cb42f28>






    

In [5]:

    

mapper = DataFrameMapper([
    ('Mission', [CountVectorizer(), TfidfTransformer()]),
])
mapper.fit(df)
X = mapper.transform(df)
y = df.Revenues


    

In [6]:

    

model = LinearRegression()
model.fit(X, y)
y_pred = model.predict(X)


    

In [7]:

    

fig = plt.figure(figsize=(8,4))
ax = plt.subplot(121)
ax.hist(y - y_pred, bins=100)
ax.set_xlabel("error")
ax.set_xscale("log")
ax = plt.subplot(122)
ax.scatter(y, y-y_pred, marker="+")
ax.set_xlabel("Revenue")
ax.set_ylabel("Error")
ax.set_xscale("log")
fig.tight_layout()
sns.despine(offset=10)


    

In [ ]: