Basic Statistics in Python

python basic stats.py

#------------------------------------------------------------------
# PROGRAM NAME: python basic stats.py
# DATE: 2/20/16
# CREATED BY: MATT BOGARD 
# PROJECT FILE:              
#----------------------------------------------------------------
# PURPOSE: BASIC STATS IN PYTHON           
#--------------------------------------------------------------- 


import pandas as pd #primary package for data manipulation


# make a data frame manually
data = {'GARST' :[150,140,145,137,141,145,149,153,157,161],
        'PIO':[160,150,146,138,142,146,150,154,158,162],
		'MYC':[137,148,151,139,143,120,115,136,130,129],
		'DEK':[150,149,145,140,144,148,152,156,160,164],
		'PLOT':[1,2,3,4,5,6,7,8,9,10],
		'BT': ['Y','Y',	'N','N','N','N','Y','N','Y','Y'],
		'RR':['Y','N','Y','N','N','N','N','Y','Y','N'],
}
yield_data = pd.DataFrame(data,columns=['GARST','PIO','MYC','DEK','PLOT','BT','RR'])

# create gmo trait  field

# set a default value
yield_data['GMO'] = 'Non-GMO    ';
                    
# check for RR traits
yield_data['GMO'][yield_data['RR']== 'Y']= 'Single Trait '
										   
# check for BT traits
yield_data['GMO'][yield_data['BT']== 'Y']= 'Single Trait '
										   
# check for double stacked traits
yield_data['GMO'][(yield_data['BT']== 'Y') & (yield_data['RR']=='Y')]= 'Stacked Trait'


#----------------------------------
# descriptive statistics
#----------------------------------

#from pandas

#descriptives
yield_data.describe()

# descriptives by group
groupby_gmo = yield_data.groupby('GMO')

groupby_gmo.mean()

#frequencies
yield_data.GMO.value_counts().sort_index()

#------------------------------
# graphics
#------------------------------

import numpy as np

import matplotlib.pyplot as plt

hist(yield_data['GARST'],bins= 3,color='green')

yield_data.plot('GARST', 'PIO', kind = 'scatter')

#-----------------------------------
# t-tests
#-----------------------------------

from scipy import stats

stats.ttest_ind(yield_data.GARST,yield_data.PIO) # output gives t-stat and pval

#welch's 2 sample (unequal var)

stats.ttest_ind(yield_data.GARST,yield_data.PIO, equal_var = False)

#run as if this were a paired ttest
stats.ttest_rel(yield_data.GARST,yield_data.PIO)

#-----------------------------------
# correlations
#-----------------------------------

# from pandas 

print( yield_data[['GARST','PIO']].corr())

#-----------------------------------
# regression
#-----------------------------------

import statsmodels.api as sm
import statsmodels.formula.api as smf

# regression using smf
results = smf.ols('GARST ~ PIO', data=yield_data).fit()

# inspect results
print results.summary()


# regression using sm
X = yield_data[['PIO']]
y = yield_data[['GARST']]

X.head()
y.head()

## fit a OLS model with intercept sales and income
X = sm.add_constant(X)
X.head()

est = sm.OLS(y, X).fit()

est.summary()