BPEstimation.py

from sklearn.ensemble import RandomForestRegressor, AdaBoostRegressor
from sklearn.linear_model import LinearRegression
from sklearn import preprocessing
from sklearn.tree import DecisionTreeRegressor


import pandas as pd
import numpy as np

Filename='features.csv'

# BP = 'dia'
BP = 'sys'

if BP=='sys':
    target = 4
if BP=='dia':
    target = 5

# Reading extracted features file    
df = pd.read_csv(Filename,error_bad_lines=False,header=None)
# Fill NANs with Means of columns
df.replace([np.inf, -np.inf],np.nan, inplace=True)
df.fillna(df.mean(),inplace=True)
df = df.reset_index(drop=True)

df.drop(df[df[4] < 90].index, inplace=True)
df.drop(df[df[5] < 40].index, inplace=True)
df.drop(df[df[10] < 50].index, inplace=True)

df.drop(df[df[4] > 200].index, inplace=True)
df.drop(df[df[5] > 100].index, inplace=True)
df.drop(df[df[10] > 140].index, inplace=True)

k_fold = 10
fold_size = int(len(df)/k_fold)
err = []
results = []
reals = []
allerrs = []
for i in range(0,k_fold):
    trainData = df.loc[0:i * fold_size]
    trainData = trainData.append(df.loc[(i+1) * fold_size:])  
    testData = df.loc[i * fold_size:(i+1) * fold_size]
    trainData = trainData[[ss for ss in range(0,np.shape(trainData)[1]-1) if (ss not in [0,1,2,3,6,7,8,9,10] or ss== target)]]
    testData = testData[[ss for ss in range(0,np.shape(testData)[1]-1) if (ss not in [0,1,2,3,6,7,8,9,10] or ss == target)]]  
  
    trainFeatures = np.array(trainData[[ss for ss in range(11,np.shape(trainData)[1]-1) if ss not in [target]]]).astype(np.float32)

    trainFeatures[trainFeatures == np.inf] = 0
    trainLabels = np.array(trainData[target]).astype(np.float32)
    
    testFeatures = np.array(testData[[ss for ss in range(11,np.shape(testData)[1]-1) if ss not in [target]]]).astype(np.float32)
    testFeatures[testFeatures == np.inf] = 0
    testLabels = np.array(testData[target]).astype(np.float32)
    
    scaler = preprocessing.StandardScaler().fit(trainFeatures)
    trainFeatures = scaler.transform(trainFeatures) 
    testFeatures = scaler.transform(testFeatures)    
    
    model = RandomForestRegressor(n_jobs=4,n_estimators=100)
    # model = AdaBoostRegressor(n_estimators=200)
    # model = LinearRegression(n_jobs=4)
    # model = DecisionTreeRegressor()
    model.fit(trainFeatures,trainLabels)
    result = model.predict(testFeatures) 
    reals.extend(testLabels)
    results.extend(result)

    err.append([i, np.mean(np.mean(np.abs(np.array(result) - np.array(testLabels))))])    
    allerrs.extend(np.abs(np.array(result) - np.array(testLabels)))
errs = [e[1] for e in err]
print('MAE: {}'.format(np.mean(allerrs)))
print('STD: {}'.format(np.std(allerrs)))
import matplotlib.pyplot as plt
import random
alldata_real = np.array(reals)
alldata_result = np.array(results)
#
corrcoef = np.corrcoef(alldata_real,alldata_result)[1][0]       
errsDetails = (np.mean(np.abs(alldata_result-alldata_real)),np.std(np.abs(alldata_result-alldata_real)))
samples = random.sample(range(len(alldata_real)), int(len(alldata_real)/10))
s45 = random.sample(range(len(alldata_real)), len(alldata_real))
fit = np.polyfit(alldata_real,alldata_result,1)
fit_fn = np.poly1d(fit) 
plt.figure()    
plt.scatter(alldata_real[[i for i in samples]],alldata_result[[i for i in samples]],s=10,c='k')
plt.plot(alldata_real, fit_fn(alldata_real), '-k')
plt.title('Regression Plot (r={})'.format(np.round(corrcoef,2)))
plt.xlabel('Target (mmHg)')
plt.ylabel('Estimate (mmHg)')
plt.savefig("./plots/RegPlot_{}.png".format(BP))
plt.show()