#!/usr/bin/env python
# coding: utf-8

# In[1]:


get_ipython().run_line_magic('matplotlib', 'inline')
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns
pd.set_option('display.width', 500)
pd.set_option('display.max_columns', 100)


# In[45]:


1+1


# #here is a heading
# 
# ##a smaller one
# 
# ###an even smaller one
# 
# ####even less important
# 
# and here is some text
# 
# hello world
# 
# http://www.google.com
# 
# [hello world](http://www.google.com)
# 
# here is an equation $E = m c^2$
# 
# $$ \alpha = \beta $$
# 
# hello **bold**, nor *emph*

# In[11]:


#also https://dl.dropboxusercontent.com/u/75194/tao-full.csv
dfsmall=pd.read_csv("https://dl.dropboxusercontent.com/u/75194/tao.csv")
dfsmall.head()


# Primary question: Can we detect an El Nino event, based on sea surface temperature? What about other observed variables?

# In[13]:


del dfsmall["Unnamed: 0"]
dfsmall.head()


# In[12]:


dfsmall.dtypes


# In[14]:


dfsmall.shape


# In[16]:


dfsmall.groupby('year').mean()#look up groupby


# In[39]:


dfsmall.groupby('year').describe()


# In[19]:


dfsmall.columns


# In[20]:


for feat in dfsmall.columns:
    print feat, dfsmall[feat].isnull().mean()


# In[27]:


for k,v in dfsmall.groupby('year'):
    for feat in dfsmall.columns:
        print k, feat, v[feat].isnull().mean()


# In[28]:


dfsmallnew=dfsmall[dfsmall['humidity'].notnull()]


# In[29]:


dfsmallnew.shape


# In[30]:


dfsmallnew=dfsmallnew[dfsmallnew['air.temp'].notnull()]
dfsmallnew=dfsmallnew[dfsmallnew['sea.surface.temp'].notnull()]
dfsmallnew.shape


# In[31]:


dfsmallnew.head()


# In[32]:


dfsmallnew.groupby('year').count()


# In[34]:


sns.pairplot(dfsmallnew, hue="year", size=2.5, alpha=0.3);#ahap happened here?
#look up http://stanford.edu/~mwaskom/software/seaborn/tutorial/axis_grids.html


# In[35]:


sns.pairplot(dfsmallnew, hue="year", size=2.5, plot_kws={'alpha':0.4});


# In[36]:


vars=list(dfsmallnew.columns)
vars


# In[38]:


sns.pairplot(dfsmallnew, hue="year", size=2.5, plot_kws={'alpha':0.4}, vars=vars[1:]);


# In[40]:


g = sns.PairGrid(dfsmallnew, vars=["sea.surface.temp", "air.temp"], hue="year")
g.map(plt.scatter)


# In[43]:


g = sns.PairGrid(dfsmallnew, vars=["sea.surface.temp", "air.temp"], hue="year")
g.map_upper(plt.scatter, alpha=0.3)
g.map_diag(sns.kdeplot, alpha=0.5)
g.map_lower(sns.kdeplot, alpha=0.5)


# In[ ]:


dfsmall.