-
Notifications
You must be signed in to change notification settings - Fork 0
/
factor_analysis.py
174 lines (150 loc) · 6.58 KB
/
factor_analysis.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
base = '/data/alstottj/Langley/'
from numpy import asarray, isnan, median, unique
from scipy.stats import skew, kruskal, ks_2samp
#from rpy2 import robjects
from rpy2.robjects.packages import importr
from rpy2.robjects.vectors import FloatVector
stats = importr('stats')
import database as db
from Helix_database import Session, database_url
session = Session()
db.create_database(database_url)
import matplotlib.pyplot as plt
from matplotlib.backends.backend_pdf import PdfPages
plots = PdfPages(base+'Langley_distributions.pdf')
import powerlaw
dependents = ['Number_of_Children', 'Parent_Child_Registration_Interval_Corrected', 'Distance_from_Parent', 'Has_Children', 'Has_Parent']
independents = ['Age', 'Gender', 'Relationship_with_Parent', 'Heard_Through_Medium', 'Same_Age_as_Parent',\
'Same_City_as_Parent', 'Same_Country_as_Parent', 'Same_Gender_as_Parent', 'Same_Relationship_to_Parent_as_They_Had_to_Their_Parent',\
'Heard_Through_Same_Medium_as_Parent', 'Has_Parent']
#robjects.r("data<-read.table('%s')"%(base+'LangleyRtable'))
#robjects.r("attach(data)")
for d in dependents:
print d
if d == 'Has_Parent':
independents = ['Age', 'Gender', 'Relationship_with_Parent', 'Heard_Through_Medium']
else:
independents = ['Age', 'Gender', 'Relationship_with_Parent', 'Heard_Through_Medium', 'Same_Age_as_Parent',\
'Same_City_as_Parent', 'Same_Country_as_Parent', 'Same_Gender_as_Parent', 'Same_Relationship_to_Parent_as_They_Had_to_Their_Parent',\
'Heard_Through_Same_Medium_as_Parent', 'Has_Parent']
for i in independents:
print i
ax = plt.subplot(1,1,1)
handles = {}
factors = session.query(db.LangleyParticipant).values(getattr(db.LangleyParticipant, i))
factors = unique([q for q in factors]).flatten()
print factors
factors = [q for q in factors if (q!=None and q!='unknown' and q!='0')]
print factors
n_factors = len(factors)
if n_factors<2:
continue
fs = []
f2s = []
Ds = []
p_krusks = []
p_KSs = []
skews = []
medians = []
for fn in range(-1, n_factors):
dist = db.LangleyDistribution()
dist.dependent = d
dist.independent = i
if fn==-1:
f='All'
data = session.query(db.LangleyParticipant).\
values(getattr(db.LangleyParticipant, d))
else:
f = factors[fn]
data = session.query(db.LangleyParticipant).\
filter(getattr(db.LangleyParticipant, i)==f).\
values(getattr(db.LangleyParticipant, d))
dist.factor = str(f)
data = [q for q in data if q[0]!=None]
data = asarray([q for q in data if ~isnan(q)]).flatten()
print str(f)+' '+str(len(data))+' cases'
if len(data)<2:
print "Skipping due to too few cases"
continue
if d=='Number_of_Children':
y, x = powerlaw.cumulative_distribution_function(data, survival=True)
if f=='All':
handles[str(f)] = ax.plot(x, y, label=str(f), linewidth=4, color='k')
else:
handles[str(f)] = ax.plot(x, y, label=str(f))
ax.set_xscale("log")
ax.set_yscale("log")
plt.title(d +' CDF, as a function of '+i)
plt.xlabel(d)
plt.ylabel('P(X)>x')
else:
if f=='All':
handles[str(f)] = plt.hist(data, bins=100, histtype='step', label=str(f), color='k')
else:
handles[str(f)] = plt.hist(data, bins=100, histtype='step', label=str(f))
ax.set_yscale("log")
plt.title(d +' PDF as a function of '+i)
plt.xlabel(d)
plt.ylabel('P(X)')
if d == 'Number_of_Children' or d == 'Has_Children':
discrete = True
else:
discrete = False
if d not in ['Has_Children', 'Has_Parent']:
fit = powerlaw.Fit(data, discrete=discrete)
R1, p1 = fit.loglikelihood_ratio('power_law', 'exponential')
R2, p2 = fit.loglikelihood_ratio('truncated_power_law', 'lognormal')
R3, p3 = fit.loglikelihood_ratio('power_law', 'lognormal')
if p1>.05:
dist.powerlaw = 0
else:
dist.powerlaw = 1
if p2<.05:
dist.powerlaw = 2
if p3<.05:
dist.powerlaw = 3
dist.mean = data.mean()
dist.median = median(data)
dist.skew = skew(data)
session.add(dist)
if f=='All':
continue
for fn2 in range(fn+1, n_factors):
f2 = factors[fn2]
data_other = session.query(db.LangleyParticipant).filter(getattr(db.LangleyParticipant, i)==f2).values(getattr(db.LangleyParticipant, d))
data_other = [q for q in data_other if q[0]!=None]
data_other = asarray([q for q in data_other if ~isnan(q)]).flatten()
if len(data_other)>2:
H, p_krusk = kruskal(data, data_other)
D, p_KS = ks_2samp(data/median(data), data_other/median(data_other))
fs.append(f)
f2s.append(f2)
Ds.append(D)
p_krusks.append(p_krusk)
p_KSs.append(p_KS)
skews.append(skew(data)-skew(data_other))
medians.append(median(data)-median(data_other))
#Adjust p values for multiple comparisons, then write
p_krusks = asarray(stats.p_adjust(FloatVector(p_krusks), method = 'holm'))
p_KSs = asarray(stats.p_adjust(FloatVector(p_KSs), method = 'holm'))
print fs
for q in range(len(fs)):
f = fs[q]
f2 = f2s[q]
comp = db.LangleyDistributionCompare()
comp.dependent = d
comp.independent = i
comp.factor1 = f
comp.factor2 = f2
comp.KW = p_krusks[q]
comp.D = Ds[q]
comp.KS = p_KSs[q]
comp.skew = skews[q]
comp.median = medians[q]
session.add(comp)
session.commit()
handles, labels = ax.get_legend_handles_labels()
ax.legend(handles[::-1], labels[::-1], loc=1)
plots.savefig()
plt.close('all')
plots.close()