Import Statement to Pip Install Statement - python-2.7

Question

Summary
I'm looking for a way to find all of the required pip install statements, with version numbers, based upon the import statements used in a script. I'll use the results to create a requirements.txt file for the environment configuration during deployment.
Issues
There is a deviation between the import name in python and the package name used by pip. Also, when using sub modules, its not immediately apparent that a pip install of the parent would be required. For instance, in our script we're doing:
import pandas as pd
#body of code
writer = pd.ExcelWriter(filename, engine='xlsxwriter')
So, if we just pip install pandas, we'll get an error that xlsxwriter needs to be installed as well.
Attempts and Research
First Attempt
#Import Packages
import pandas as pd
from sklearn import linear_model
#Create an Array of the Package names
imports = ['pandas','sklearn']
#find installed package version on local machine for each imported package
import pip #needed to use the pip functions
for i in pip.get_installed_distributions(local_only=True):
for ii in imports:
if str(i).find(str(ii)) != -1:
print(str(i))
#pandas 0.18.1 is found but scikit-learn 0.17.1 is not
Research
The below from an SO post appears to list out all of the import names, but they after running the second answer in the post, the import names I would expect to see such as 'sklearn' were missing.
print(list(item[1] for item in pkgutil.iter_modules()))