Question

我正在寻找一种计算因素之间得分变化的方法（例如，治疗前后的问卷得分）。我想弄清楚参与者的百分比提高了，而事前和事后之间没有改善。

我看过一些dplyr解决方案，但我认为我缺少其中的一行代码，但不确定。

    ID<-c("aaa","bbb","ccc","ddd","eee","fff", "ggg","aaa","bbb","ccc","ddd","eee","fff", "ggg")
    Score<-sample(40,14)
    Pre_Post<-c(1,1,1,1,1,1,1,2,2,2,2,2,2,2)
    df<-cbind(ID, Pre_Post, Score)
    df$Score<-as.numeric(df$Score)
    df<-as.data.frame(df)


    #what I have tried
    df2<-df%>%
    group_by(ID, Pre_post)
    mutate(Pct_change=mutate(Score/lead(Score)*100))

但是我收到错误消息。同样，我也不认为代码是正确的开始。

预期结果：- 我要实现的是获取已改善的ID的百分比。因此，以我提供的模拟示例为例，从Pre到Post仅改善了ID的42.86％，而在Pre和Post之间实际上恶化了57.14％。

任何建议都将受到欢迎：）

Answer 1

您有几种错别字，这就是为什么您会收到错误消息。

您可以执行以下操作来同时获得新旧分数：

#!/usr/bin/env python3

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.colors
from mpl_toolkits.axes_grid1 import make_axes_locatable, axes_size 

# border limits for plots
lowlim = 0.43
uplim = 0.52
nx = 10


kernels = ['cubic_spline', 'quintic_spline',
        'wendland_C2', 'wendland_C4', 'wendland_C6']




#========================
def main():
#========================


    eta_facts = [ 0, 1, 2, 3, 4, 5 ]

    nrows = len(eta_facts)
    ncols = len(kernels)

    Ay_list = [[None for c in range(ncols)] for r in range(nrows)]



    #--------------------------------
    # Loop and compute As
    #--------------------------------
    dx = (uplim - lowlim)/nx

    for row, eta in enumerate(eta_facts):

        for col, kernel in enumerate(kernels):

            A = np.zeros((nx, nx), dtype=np.float)

            for i in range(nx):
                for j in range(nx):
                    A[j,i] = row + np.random.random()/10 # not a typo: need A[j,i] for imshow

            Ay_list[row][col] = A




    #------------------------------------
    # Now plot it
    #------------------------------------

    fig = plt.figure(figsize=(3.5*ncols, 3.5*nrows))

    axrows = []
    i = 1
    for r in range(nrows):
        axcols = []
        for c in range(ncols):
            if r > 0:
                if c > 0:
                    axcols.append(fig.add_subplot(nrows, ncols, i, 
                            aspect='equal', sharex=axrows[r-1][c], sharey=axcols[c-1]))
                else:
                    axcols.append(fig.add_subplot(nrows, ncols, i, 
                            aspect='equal', sharex=axrows[r-1][c]))
            else:
                if c > 0:
                    axcols.append(fig.add_subplot(nrows, ncols, i, 
                            aspect='equal', sharey=axcols[c-1]))
                else:
                    axcols.append(fig.add_subplot(nrows, ncols, i, aspect='equal'))
            i+=1
        axrows.append(axcols)



    cmap = 'YlGnBu_r'

    lw = 2


    for row in range(nrows):
        axcols = axrows[row]

        minval = min([np.min(Ay_list[row][c]) for c in range(ncols)])
        maxval = max([np.max(Ay_list[row][c]) for c in range(ncols)])


        for col, ax in enumerate(axcols):

            im = ax.imshow(Ay_list[row][col], origin='lower', 
                vmin=minval, vmax=maxval, cmap=cmap,
                extent=(lowlim, uplim, lowlim, uplim),
                #  norm=matplotlib.colors.SymLogNorm(1e-3),
                zorder=1)

            # only plot colorbar for last column

            if col==len(kernels)-1:
                divider = make_axes_locatable(ax)
                cax = divider.append_axes("right", size="5%", pad=0.15)
                fig.colorbar(im, cax=cax)


            ax.set_xlim((lowlim,uplim))
            ax.set_ylim((lowlim,uplim))


            # cosmetics
            if col > 0:
                left = False
            else:
                left = True
            if row == len(eta_facts)-1 :
                bottom = True
            else:
                bottom = False

            ax.tick_params(
                axis='both',        # changes apply to the x-axis
                which='both',       # both major and minor ticks are affected
                bottom=bottom,      # ticks along the bottom edge are off
                top=False,          # ticks along the top edge are off
                left=left,          # ticks along the left edge are off
                right=False,        # ticks along the rigt edge are off
                labelbottom=bottom, # labels along the bottom edge are off
                labeltop=False,     # labels along the top edge are off
                labelleft=left,     # labels along the left edge are off
                labelright=False)   # labels along the right edge are off


            if row==0:
                ax.set_title(kernels[col] + ' kernel', fontsize=14)
            if col==0:
                ax.set_ylabel(r"$\eta = $ "+str(eta_facts[row])+r"$\eta_0$")


    fig.suptitle(r"Some title", fontsize=18)
    plt.tight_layout(rect=(0, 0, 1, 0.97))
    plt.subplots_adjust(wspace=0, hspace=0)
    plt.savefig('for_stackexchange.png', dpi=150)
    plt.close()

    print('finished.')

    return





if __name__ == '__main__':
    main()

要获取改进数量，您必须先将library(tidyverse) df %>% spread(Pre_Post, Score) %>% rename(Score_pre = `1`, Score_post = `2`) ID Score_pre Score_post 1 aaa 19 24 2 bbb 39 35 3 ccc 2 29 4 ddd 38 15 5 eee 36 9 6 fff 23 10 7 ggg 21 27转换为数字：

Score

Answer 2

另一种选择dplyr的选项是假设您总是有两个值，其中Pre为1，Post为2，这将是group_by ID并减去第二个值和第一个值，并计算正负值的比率。

library(dplyr)

df %>%
  arrange(ID, Pre_Post) %>%
  group_by(ID) %>%
  summarise(val = Score[2] - Score[1]) %>%
  summarise(total_pos = sum(val > 0)/n(), 
            total_neg = sum(val < 0)/ n())

# A tibble: 1 x 2
#  total_pos total_neg
#      <dbl>     <dbl>
#1     0.429     0.571

数据

ID <- c("aaa","bbb","ccc","ddd","eee","fff", "ggg","aaa","bbb",
      "ccc","ddd","eee","fff", "ggg")
Score <- sample(40,14)
Pre_Post <- c(1,1,1,1,1,1,1,2,2,2,2,2,2,2)
df <- data.frame(ID, Pre_Post, Score)

计算因素之间的增加和减少值的百分比

2 个答案: