-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathDigitalEpidemiology.py
executable file
·145 lines (109 loc) · 4.8 KB
/
DigitalEpidemiology.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
# -*- coding: utf-8 -*-
"""
Created on Wed Nov 6 10:34:09 2019
@author: aless
"""
"""
Project work of the course DIGITAL EPIDEMIOLOGY
AUTHORS: Emanuela Iovino, Alessia Angeli, Paola Dimartino, Erika Gardini
GROUND TRUTH: depressione (crude prevalence by region in usa)
DIGITAL PROXY DATA: antidepressivo, autostima, barbiturato, creeper, depressione, facebook, fluoxetina, incel, ketamina, paroxetina, prozac, sertralina
YEARS: 2016, 2017, 2018
"""
import matplotlib
matplotlib.use('Agg')
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.linear_model import LinearRegression
import seaborn as sns
#Read from csv
data2016 = pd.read_csv("data2016.csv")
data2016 = data2016.fillna(0)
data2017 = pd.read_csv("data2017.csv")
data2017 = data2017.fillna(0)
data2018 = pd.read_csv("data2018.csv")
data2018 = data2018.fillna(0)
print("Correlation matrix gt vs proxy data - 2016")
corr2016 = data2016.corr()
ax = sns.heatmap(corr2016, vmin=-1, vmax=1, center=0, cmap=sns.diverging_palette(20,220,n=200), square=True)
ax.set_xticklabels(data2016.columns[1:],rotation=45, horizontalalignment='right')
plt.show()
print("Correlation matrix gt vs proxy data - 2017")
corr2017 = data2017.corr()
ax = sns.heatmap(corr2017, vmin=-1, vmax=1, center=0, cmap=sns.diverging_palette(20,220,n=200), square=True)
ax.set_xticklabels(data2017.columns[1:],rotation=45, horizontalalignment='right')
plt.show()
print("Correlation matrix gt vs proxy data - 2018")
corr2018 = data2018.corr()
ax = sns.heatmap(corr2018, vmin=-1, vmax=1, center=0, cmap=sns.diverging_palette(20,220,n=200), square=True)
ax.set_xticklabels(data2018.columns[1:],rotation=45, horizontalalignment='right')
plt.show()
print("We can observe that the correlation matrices are similar across different years.")
print("")
def fit_the_model(x_train, y_train):
model = LinearRegression()
model.fit(x_train, y_train)
print("r_sq ", str(model.score(x_train, y_train)))
return model
def eval_model(x_test, y_test, model):
y_pred = model.predict(x_test)
error = (np.abs(y_test - y_pred) / y_test) * 100
result = pd.DataFrame({'Actual': y_test.flatten(), 'Predicted': y_pred.flatten(), 'Error(%)': error})
print(result)
return error
file_names = ["data2016", "data2017", "data2018"]
global_error = []
mean_error_per_year = []
global_error_income = []
mean_error_per_year_income = []
### Cross validation: we repeat the experiment three times, using as training set the data from two years
### and as test set the remaining data
for el in file_names:
test = pd.read_csv(el + ".csv")
test = data2016.fillna(0)
train_name = list(filter(lambda x: (x != el), file_names))
train1 = pd.read_csv(train_name[0] + ".csv")
train1 = train1.fillna(0)
train2 = pd.read_csv(train_name[1] + ".csv")
train2 = train2.fillna(0)
train1 = train1.values
train2 = train2.values
train = np.concatenate((train1, train2))
x_train = train[:, 2:-1]
x_train_income = train[:, 2:]
y_train = train[:, 1]
test = test.values
x_test = test[:, 2:-1]
x_test_income = test[:, 2:]
y_test = test[:, 1]
print("TRAIN: " + train_name[0] + " " + train_name[1])
print("TEST: " + el)
print("Compute linear regressor without income...")
model = fit_the_model(x_train, y_train)
print("Show error of the model")
error = eval_model(x_train, y_train, model)
global_error.append(error)
mean_error_per_year.append(np.mean(error))
print("Compute linear regressor with income...")
model = fit_the_model(x_train_income, y_train)
print("Show error of the model")
error = eval_model(x_train_income, y_train, model)
global_error_income.append(error)
mean_error_per_year_income.append(np.mean(error))
global_mean_error = np.array(global_error)
global_mean_error = np.mean(global_mean_error)
mean_error_per_year = np.array(mean_error_per_year)
print("Global mean error without income " + str(global_mean_error))
print("Mean error per year without income " + str(mean_error_per_year))
print("\n")
global_mean_error_income = np.array(global_error_income)
global_mean_error_income = np.mean(global_mean_error_income)
mean_error_per_year_income = np.array(mean_error_per_year_income)
print("Global mean error with income " + str(global_mean_error_income))
print("Mean error per year with income " + str(mean_error_per_year_income))
print("")
print("We can observe that adding the information about the householding income the performance of the model is not improved.")
print("")
if(global_mean_error!=0 and len(mean_error_per_year)!=0 and global_mean_error_income!=0 and len(mean_error_per_year_income)!=0):
print("Script correctly executed")