熊猫：应用搜索功能

Question

I am trying to reproduce some neat forest plots on my own data. However, I'm stuck at this function, and can't for the life of me figure out what its supposed to do.

I am trying to make the following code work on my data:

def create_smry(trc, data, pname='subject'):
    ''' Conv fn: create trace summary for sorted forestplot '''

    dfsm = pm.df_summary(trc).reset_index()
    dfsm.rename(columns={'index':'featval'}, inplace=True)

    print(dfsm.head(n=5))

    dfsm = dfsm.loc[dfsm['featval'].apply(
        lambda x: re.search('{}__[0-9]+'.format(pname), x) is not None)]

    dfsm.set_index(dfs[pname].unique(), inplace=True)
    dfsm.sort('mean', ascending=True, inplace=True)
    dfsm['ypos'] = np.arange(len(dfsm))

    return dfsm

where the print returns:

  featval      mean        sd  mc_error   hpd_2.5  hpd_97.5
0    mu_a -0.008913  0.011715  0.000613 -0.029139  0.014329
1    mu_b  0.003252  0.000271  0.000015  0.002698  0.003765
2    a__0 -0.065255  0.024315  0.001168 -0.113708 -0.018885
3    a__1 -0.081748  0.023247  0.001114 -0.124560 -0.036777
4    a__2  0.025326  0.021661  0.001024 -0.019744  0.065263

The error:

---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
<ipython-input-125-2465af1d68b8> in <module>()
----> 1 dfsm_unpl_mfr = create_smry(hierarchical_trace[-333:], data, 'subject')
      2 custom_forestplot(dfsm_unpl_mfr)

<ipython-input-123-5f6828d6cf8e> in create_smry(trc, data, pname)
      8 
      9     dfsm = dfsm.loc[dfsm['featval'].apply(
---> 10         lambda x: re.search('{}__[0-9]+'.format(pname), x) is not None)]
     11 
     12     dfsm.set_index(dfs[pname].unique(), inplace=True)

~/anaconda/envs/py35/lib/python3.5/site-packages/pandas/core/series.py in apply(self, func, convert_dtype, args, **kwds)
   2353             else:
   2354                 values = self.asobject
-> 2355                 mapped = lib.map_infer(values, f, convert=convert_dtype)
   2356 
   2357         if len(mapped) and isinstance(mapped[0], Series):

pandas/_libs/src/inference.pyx in pandas._libs.lib.map_infer (pandas/_libs/lib.c:66645)()

<ipython-input-123-5f6828d6cf8e> in <lambda>(x)
      8 
      9     dfsm = dfsm.loc[dfsm['featval'].apply(
---> 10         lambda x: re.search('{}__[0-9]+'.format(pname), x) is not None)]
     11 
     12     dfsm.set_index(dfs[pname].unique(), inplace=True)

NameError: name 're' is not defined

1. I cant figure out what re.search is, since re is not a df.
2. What does {}__[0-9]+ mean in this context?

Since the input is quite complicated, I can't provide a minimal working example.

After import of regex:

import re

def create_smry(trc, data, pname='subject'):
    ''' Conv fn: create trace summary for sorted forestplot '''

    dfsm = pm.df_summary(trc).reset_index()
    dfsm.rename(columns={'index':'featval'}, inplace=True)

    print(dfsm.head(n=10))

    dfsm = dfsm.loc[dfsm['featval'].apply(
        lambda x: re.search('{}__[0-90]+'.format(pname), x) is not None)]

    print(dfsm.head(n=10))

    dfsm.set_index(data[pname].unique(), inplace=True)
    dfsm.sort_values('mean', ascending=True, inplace=True)
    dfsm['ypos'] = np.arange(len(dfsm))

    print(dfsm.head(n=15))

    return dfsm

which returns

  featval      mean        sd  mc_error   hpd_2.5  hpd_97.5
0   b0_mu -0.022521  0.010266  0.000597 -0.042222 -0.003072
1   b1_mu  0.003220  0.000256  0.000014  0.002742  0.003700
2   b2_mu  0.024366  0.005288  0.000292  0.014786  0.035139
3   b3_mu  0.008563  0.004393  0.000243  0.000634  0.017385
4   b0__0 -0.078060  0.025093  0.001208 -0.121480 -0.024921
5   b0__1 -0.097636  0.024500  0.001413 -0.144801 -0.052600
6   b0__2  0.009216  0.024381  0.001229 -0.038927  0.052254
7   b0__3  0.024541  0.025525  0.001399 -0.025824  0.070295
8   b0__4 -0.069331  0.020887  0.001057 -0.106392 -0.024169
9   b0__5 -0.065629  0.024787  0.001178 -0.111582 -0.019849
Empty DataFrame
Columns: [featval, mean, sd, mc_error, hpd_2.5, hpd_97.5]
Index: []

If I block out the re.search and simply plot (also dont try to change the index, I get a plot:

However, re.search was not employed correctly, so all y-values from trc fra plotted.

EDIT : Ended up using

dfsm['featidx'] = dfsm['featval'].apply(lambda x: any(pd.Series(x).str.contains(feat)))

since I could not figure out regex.

Answer 1

I cant figure out what re.search is, since re is not a df.

re is a regex library to perform regex actions on a string. You need to call import re in the header or python file to use it.

What does {}__[0-9]+ mean in this context?

It is a regex pattern, re.search says, scan through string looking for a location where this regular expression ( {}__[0-9]+ ) produces a match, and return a corresponding match object .

For more info about the library : ' Regex Documentation '