Related
I aim to compare the restricted mean survival time between the two treatment groups in the Anderson dataset
Anderson dataset
Here is the structure of my data frame:
'data.frame': 42 obs. of 5 variables:
$ survt : num 19 17 13 11 10 10 9 7 6 6 ...
$ status: num 0 0 1 0 0 1 0 1 0 1 ...
$ sex : Factor w/ 2 levels "female","male": 1 1 1 1 1 1 1 1 1 1 ...
$ logwbc: 'labelled' num 2.05 2.16 2.88 2.6 2.7 2.96 2.8 4.43 3.2 2.31 ...
..- attr(*, "label")= Named chr "log WBC"
.. ..- attr(*, "names")= chr "logwbc"
$ rx : Factor w/ 2 levels "New treatment",..: 1 1 1 1 1 1 1 1 1 1 ...
..- attr(*, "label")= Named chr "Treatment"
.. ..- attr(*, "names")= chr "rx"
- attr(*, "codepage")= int 65001
I used the following code to compare the restricted mean survival time between the two treatment groups ("New treatment" vs. "Standard treatment):
time <- anderson$survt
status <- anderson$status
arm <- anderson$rx
rmst2(time, status, arm )
I get the following error:
Error in rmst2(time, status, arm) : object 'NOTE' not found
In addition: Warning messages:
1: In max(tt) : no non-missing arguments to max; returning -Inf
2: In min(ss[tt == tt0max]) :
no non-missing arguments to min; returning Inf
3: In max(tt) : no non-missing arguments to max; returning -Inf
4: In min(ss[tt == tt1max]) :
no non-missing arguments to min; returning Inf
Thanks
I converted the sex and rx variables from factor to numeric and the function worked.
I have dataframe with columns A,B and flag. I want to calculate mean of 2 values before flag change from 0 to 1 , and record value when flag change from 0 to 1 and record value when flag changes from 1 to 0.
# Input dataframe
df=pd.DataFrame({'A':[1,3,4,7,8,11,1,15,20,15,16,87],
'B':[1,3,4,6,8,11,1,19,20,15,16,87],
'flag':[0,0,0,0,1,1,1,0,0,0,0,0]})
# Expected output
df_out=df=pd.DataFrame({'A_mean_before_flag_change':[5.5],
'B_mean_before_flag_change':[5],
'A_value_before_change_flag':[7],
'B_value_before_change_flag':[6]})
I try to create more general solution:
df=pd.DataFrame({'A':[1,3,4,7,8,11,1,15,20,15,16,87],
'B':[1,3,4,6,8,11,1,19,20,15,16,87],
'flag':[0,0,0,0,1,1,1,0,0,1,0,1]})
print (df)
A B flag
0 1 1 0
1 3 3 0
2 4 4 0
3 7 6 0
4 8 8 1
5 11 11 1
6 1 1 1
7 15 19 0
8 20 20 0
9 15 15 1
10 16 16 0
11 87 87 1
First create groups by mask for 0 with next 1 values of flag:
m1 = df['flag'].eq(0) & df['flag'].shift(-1).eq(1)
df['g'] = m1.iloc[::-1].cumsum()
print (df)
A B flag g
0 1 1 0 3
1 3 3 0 3
2 4 4 0 3
3 7 6 0 3
4 8 8 1 2
5 11 11 1 2
6 1 1 1 2
7 15 19 0 2
8 20 20 0 2
9 15 15 1 1
10 16 16 0 1
11 87 87 1 0
then filter out groups with size less like N:
N = 4
df1 = df[df['g'].map(df['g'].value_counts()).ge(N)].copy()
print (df1)
A B flag g
0 1 1 0 3
1 3 3 0 3
2 4 4 0 3
3 7 6 0 3
4 8 8 1 2
5 11 11 1 2
6 1 1 1 2
7 15 19 0 2
8 20 20 0 2
Filter last N rows:
df2 = df1.groupby('g').tail(N)
And aggregate last with mean:
d = {'mean':'_mean_before_flag_change', 'last': '_value_before_change_flag'}
df3 = df2.groupby('g')['A','B'].agg(['mean','last']).sort_index(axis=1, level=1).rename(columns=d)
df3.columns = df3.columns.map(''.join)
print (df3)
A_value_before_change_flag B_value_before_change_flag \
g
2 20 20
3 7 6
A_mean_before_flag_change B_mean_before_flag_change
g
2 11.75 12.75
3 3.75 3.50
I'm assuming that this needs to work for cases with more than one rising edge and that the consecutive values and averages get appended to the output lists:
# the first step is to extract the rising and falling edges using diff(), identify sections and length
df['flag_diff'] = df.flag.diff().fillna(0)
df['flag_sections'] = (df.flag_diff != 0).cumsum()
df['flag_sum'] = df.flag.groupby(df.flag_sections).transform('sum')
# then you can get the relevant indices by checking for the rising edges
rising_edges = df.index[df.flag_diff==1.0]
val_indices = [i-1 for i in rising_edges]
avg_indices = [(i-2,i-1) for i in rising_edges]
# and finally iterate over the relevant sections
df_out = pd.DataFrame()
df_out['A_mean_before_flag_change'] = [df.A.loc[tpl[0]:tpl[1]].mean() for tpl in avg_indices]
df_out['B_mean_before_flag_change'] = [df.B.loc[tpl[0]:tpl[1]].mean() for tpl in avg_indices]
df_out['A_value_before_change_flag'] = [df.A.loc[idx] for idx in val_indices]
df_out['B_value_before_change_flag'] = [df.B.loc[idx] for idx in val_indices]
df_out['length'] = [df.flag_sum.loc[idx] for idx in rising_edges]
df_out.index = rising_edges
I have two questions for calibrate with cph function.
My data have 5 independent variables(from BMI to RT), and 2 dependent variables (time, event).
> head(data)
BMI Taxanes Surgery LND RT Event Time
1 19 0 0 2 5 0 98
2 20 0 0 3 3 0 97
3 21 0 0 8 2 0 17
4 18 0 0 1 3 0 35
5 20 1 0 3 1 0 27
6 20 1 0 2 3 1 2
> str(data)
$ BMI : num 19 20 21 18 20 20 20 ...
$ Taxanes: int 0 0 0 0 1 1 1 0 0 0 ...
$ Surgery: num 0 0 0 0 0 0 1 0 0 0 ...
$ LND : int 2 3 8 1 3 2 2 2 5 2 ...
$ RT : Factor w/ 7 levels "0","1","2","3",..: 5 3 2 3 1 3 ...
$ Event : int 0 0 0 0 0 1 0 0 0 0 ...
$ Time : num 98 97 17 35 27 2 22 ...
(1) With this data, I did survival analysis with cph model. And I want to make a calibration plot using this data. But I got an error which "Error in x(x) : argument "y" is missing, with no default". I was finding lots of material. But I don't know the reason for this error. Even if I found the calibrate function in web, But I can't find for the element 'y'. please help me for this question.
> ddist <- datadist(data)
> options(datadist='ddist')
>
> fit = cph(Surv(Time,Event) ~ BMI + Surgery + Taxanes + RT + LND, data=data, x=TRUE, y=TRUE, surv=TRUE, dxy=TRUE, time.inc=36)
> plot(calibrate(fit))
Using Cox survival estimates at 36 Days
**Error in x(x) : argument "y" is missing, with no default**
(2) Eventually I want to do external validation for this cph model(fit).
If new data name is kind of dat2 (which has the same variable with data), then what is the observed and predicted survival? I know that the predicted value calculate like this code
val<-val.surv(fit, newdata=dat2, S=Surv(dat2$Time,dat2$Event))
But how I get a actual(observed) survival in new data(dat2)? Please help for this problem. Thank you so much in advance!
I'm trying to check the cartesian distance between each set of points in one dataframe to sets of scattered points in another dataframe, to see if the input gets above a threshold 'distance' of my checking points.
I have this working with nested for loops, but is painfully slow (~7 mins for 40k input rows, each checked vs ~180 other rows, + some overhead operations).
Here is what I'm attempting in vectorialized format - 'for every pair of points (a,b) from df1, if the distance to ANY point (d,e) from df2 is > threshold, print "yes" into df1.c, next to input points.
..but I'm getting unexpected behavior from this. With given data, all but one distances are > 1, but only df1.1c is getting 'yes'.
Thanks for any ideas - the problem is probably in the 'df1.loc...' line:
import numpy as np
from pandas import DataFrame
inp1 = [{'a':1, 'b':2, 'c':0}, {'a':1,'b':3,'c':0}, {'a':0,'b':3,'c':0}]
df1 = DataFrame(inp1)
inp2 = [{'d':2, 'e':0}, {'d':0,'e':3}, {'d':0,'e':4}]
df2 = DataFrame(inp2)
threshold = 1
df1.loc[np.sqrt((df1.a - df2.d) ** 2 + (df1.b - df2.e) ** 2) > threshold, 'c'] = "yes"
print(df1)
print(df2)
a b c
0 1 2 yes
1 1 3 0
2 0 3 0
d e
0 2 0
1 0 3
2 0 4
Here is an idea to help you to start...
Source DFs:
In [170]: df1
Out[170]:
c x y
0 0 1 2
1 0 1 3
2 0 0 3
In [171]: df2
Out[171]:
x y
0 2 0
1 0 3
2 0 4
Helper DF with cartesian product:
In [172]: x = df1[['x','y']] \
.reset_index() \
.assign(k=0).merge(df2.assign(k=0).reset_index(),
on='k', suffixes=['1','2']) \
.drop('k',1)
In [173]: x
Out[173]:
index1 x1 y1 index2 x2 y2
0 0 1 2 0 2 0
1 0 1 2 1 0 3
2 0 1 2 2 0 4
3 1 1 3 0 2 0
4 1 1 3 1 0 3
5 1 1 3 2 0 4
6 2 0 3 0 2 0
7 2 0 3 1 0 3
8 2 0 3 2 0 4
now we can calculate the distance:
In [169]: x.eval("D=sqrt((x1 - x2)**2 + (y1 - y2)**2)", inplace=False)
Out[169]:
index1 x1 y1 index2 x2 y2 D
0 0 1 2 0 2 0 2.236068
1 0 1 2 1 0 3 1.414214
2 0 1 2 2 0 4 2.236068
3 1 1 3 0 2 0 3.162278
4 1 1 3 1 0 3 1.000000
5 1 1 3 2 0 4 1.414214
6 2 0 3 0 2 0 3.605551
7 2 0 3 1 0 3 0.000000
8 2 0 3 2 0 4 1.000000
or filter:
In [175]: x.query("sqrt((x1 - x2)**2 + (y1 - y2)**2) > #threshold")
Out[175]:
index1 x1 y1 index2 x2 y2
0 0 1 2 0 2 0
1 0 1 2 1 0 3
2 0 1 2 2 0 4
3 1 1 3 0 2 0
5 1 1 3 2 0 4
6 2 0 3 0 2 0
Try using scipy implementation, it is surprisingly fast
scipy.spatial.distance.pdist
https://docs.scipy.org/doc/scipy/reference/generated/scipy.spatial.distance.pdist.html
or
scipy.spatial.distance_matrix
https://docs.scipy.org/doc/scipy-0.19.1/reference/generated/scipy.spatial.distance_matrix.html
I'm rather new at python.
I try to have a cumulative sum for each client to see the consequential months of inactivity (flag: 1 or 0). The cumulative sum of the 1's need therefore to be reset when we have a 0. The reset need to happen as well when we have a new client. See below with example where a is the column of clients and b are the dates.
After some research, I found the question 'Cumsum reset at NaN' and 'In Python Pandas using cumsum with groupby'. I assume that I kind of need to put them together.
Adapting the code of 'Cumsum reset at NaN' to the reset towards 0, is successful:
cumsum = v.cumsum().fillna(method='pad')
reset = -cumsum[v.isnull() !=0].diff().fillna(cumsum)
result = v.where(v.notnull(), reset).cumsum()
However, I don't succeed at adding a groupby. My count just goes on...
So, a dataset would be like this:
import pandas as pd
df = pd.DataFrame({'a' : [1,1,1,1,1,1,1,2,2,2,2,2,2,2],
'b' : [1/15,2/15,3/15,4/15,5/15,6/15,1/15,2/15,3/15,4/15,5/15,6/15],
'c' : [1,0,1,0,1,1,0,1,1,0,1,1,1,1]})
this should result in a dataframe with the columns a, b, c and d with
'd' : [1,0,1,0,1,2,0,1,2,0,1,2,3,4]
Please note that I have a very large dataset, so calculation time is really important.
Thank you for helping me
Use groupby.apply and cumsum after finding contiguous values in the groups. Then groupby.cumcount to get the integer counting upto each contiguous value and add 1 later.
Multiply with the original row to create the AND logic cancelling all zeros and only considering positive values.
df['d'] = df.groupby('a')['c'] \
.apply(lambda x: x * (x.groupby((x != x.shift()).cumsum()).cumcount() + 1))
print(df['d'])
0 1
1 0
2 1
3 0
4 1
5 2
6 0
7 1
8 2
9 0
10 1
11 2
12 3
13 4
Name: d, dtype: int64
Another way of doing would be to apply a function after series.expanding on the groupby object which basically computes values on the series starting from the first index upto that current index.
Use reduce later to apply function of two args cumulatively to the items of iterable so as to reduce it to a single value.
from functools import reduce
df.groupby('a')['c'].expanding() \
.apply(lambda i: reduce(lambda x, y: x+1 if y==1 else 0, i, 0))
a
1 0 1.0
1 0.0
2 1.0
3 0.0
4 1.0
5 2.0
6 0.0
2 7 1.0
8 2.0
9 0.0
10 1.0
11 2.0
12 3.0
13 4.0
Name: c, dtype: float64
Timings:
%%timeit
df.groupby('a')['c'].apply(lambda x: x * (x.groupby((x != x.shift()).cumsum()).cumcount() + 1))
100 loops, best of 3: 3.35 ms per loop
%%timeit
df.groupby('a')['c'].expanding().apply(lambda s: reduce(lambda x, y: x+1 if y==1 else 0, s, 0))
1000 loops, best of 3: 1.63 ms per loop
I think you need custom function with groupby:
#change row with index 6 to 1 for better testing
df = pd.DataFrame({'a' : [1,1,1,1,1,1,1,2,2,2,2,2,2,2],
'b' : [1/15,2/15,3/15,4/15,5/15,6/15,1/15,2/15,3/15,4/15,5/15,6/15,7/15,8/15],
'c' : [1,0,1,0,1,1,1,1,1,0,1,1,1,1],
'd' : [1,0,1,0,1,2,3,1,2,0,1,2,3,4]})
print (df)
a b c d
0 1 0.066667 1 1
1 1 0.133333 0 0
2 1 0.200000 1 1
3 1 0.266667 0 0
4 1 0.333333 1 1
5 1 0.400000 1 2
6 1 0.066667 1 3
7 2 0.133333 1 1
8 2 0.200000 1 2
9 2 0.266667 0 0
10 2 0.333333 1 1
11 2 0.400000 1 2
12 2 0.466667 1 3
13 2 0.533333 1 4
def f(x):
x.ix[x.c == 1, 'e'] = 1
a = x.e.notnull()
x.e = a.cumsum()-a.cumsum().where(~a).ffill().fillna(0).astype(int)
return (x)
print (df.groupby('a').apply(f))
a b c d e
0 1 0.066667 1 1 1
1 1 0.133333 0 0 0
2 1 0.200000 1 1 1
3 1 0.266667 0 0 0
4 1 0.333333 1 1 1
5 1 0.400000 1 2 2
6 1 0.066667 1 3 3
7 2 0.133333 1 1 1
8 2 0.200000 1 2 2
9 2 0.266667 0 0 0
10 2 0.333333 1 1 1
11 2 0.400000 1 2 2
12 2 0.466667 1 3 3
13 2 0.533333 1 4 4