Cbr分析

import pandas as pd import numpy as np def mean_daily_features(Data): ind = 0 cols = ['SubjID','Duration_Mean','Duration_Median','Wear Fraction','NewBrace'] Features = pd.DataFrame(data=[],columns=cols) #create dataframe with daily estimates dailyestimates = pd.DataFrame() for newbrace in Data.NewBrace.unique(): for s in Data.SubjID.unique(): d = Data[(Data.SubjID==s)&(Data.NewBrace==newbrace)] meandur = d['Duration'].mean() mediandur = d['Duration'].median() wearfrac = d['Wear Frac'].mean() #average across all destinations and visits sixmwt = np.mean(d['6mwt']) tenmwt = np.mean(d['10mwt_ss']) cadence = np.nanmean(d['Cadence']) # UsageFrac = np.sum(d['Wear Frac']*d['Duration'])/np.sum(d['Duration']) stepsHr = np.nanmean(d['Steps per Hour']) age = np.unique(d['Age']) BaselineYrs = np.unique(d['BaselineYrs']) #daily averages dailydur = []; dailysteps = []; dailycadence = []; dailywf = [] for dates in d['Date'].unique(): dailysteps.append(d.loc[d['Date']==dates,'Steps'].sum()) #total daily steps dailycadence.append(d.loc[d['Date']==dates,'Cadence'].mean()) #daily cadence dailydur.append(d.loc[d['Date']==dates,'Duration'].sum()) #daily time dailywf.append(d.loc[d['Date']==dates,'Wear Frac'].mean()) #mean daily wf dailyestimates = pd.concat((dailyestimates, pd.DataFrame({'Steps':dailysteps, 'Cadence':dailycadence, 'Time Community':dailydur, 'Wear Fraction':dailywf, 'NewBrace':newbrace}, index=np.tile(s, len(dailysteps)))), axis=0) #daily averages meandailysteps = np.mean(dailysteps) meandailycadence = np.nanmean(dailycadence) meandailydur = np.mean(dailydur) meandailywf = np.mean(dailywf) Features_ = pd.DataFrame({cols[0]:s, cols[1]:meandur, cols[2]:mediandur, cols[3]:meandailywf, cols[4]:newbrace, '6mwt':sixmwt, '10mwt_ss':tenmwt, 'Steps':meandailysteps, 'Cadence':meandailycadence, 'Daily Duration':meandailydur, 'Age':age, 'BaselineYrs':BaselineYrs},index=[ind]) Features = pd.concat([Features,Features_]) ind+=1 return Features, dailyestimates