Python中的ANOVA

Question

My csv file contains discrete and continuous variables and I would like to find the "equation of model" which explains my continuous variable (a) according to my discrete variables(x,y,z); ->a=f(x,y,z).The problem is that i'm trying this code but it fails. When I want to see the result of the print([f_value, p_value] stats.f_oneway = (x, y, z)), I get [nan, nan]. Extract from my code:

from numpy import (genfromtxt,hstack,arange)
#Pr linear regression
from scipy import stats
import scipy
#Pr ANOVA
from statsmodels.stats.multicomp import (pairwise_tukeyhsd,MultiComparison)

from pylab import savefig
from matplotlib.pyplot import (figure,setp)

fname="G:/table.csv"
my_data = genfromtxt(fname,delimiter=',')

#Transformation of file into table

x= my_data[:,3]
y= my_data[:,4]
z= my_data[:,6]


#one way anova
[f_value, p_value] = stats.f_oneway(x, y, z)

I want estimate the model R=u+f(x,y,z)+ e where R (continuous variable), u (constant), e (measurement error). I want have the coefficients for how the discrete variables affect my continuous outcome.

Answer 1

My best guess is that your data contains NaN values, or bad values. You could try to detect is with following piece of code:

for i in range(1,len(x)):
    (f_val,p_val) = stats.f_oneway(x[:i],y[:i],z[:i])
    if numpy.isnan(f_val) or numpy.isnan(p_val):
        print i-1,x[i-1],y[i-1],z[i-1],f_val,p_val