scripting

Anything about scripting – Python, R, Javascript, etc

ZBP Data in a Notebook

Examples of using the Census Bureau’s API with Python

At the end of my book I briefly illustrate how the Census Bureau’s API works using Python. I’ll expand on that in this post; we’ll pull data from the Population Estimates Program, transform it, and create a chart using Python with Pandas in a Notebook. I’ll conclude with an additional example using the ZIP Code Business Patterns.

The Census Bureau has dedicated API pages for each dataset (decennial, acs, pop estimates, and more), and you need to familiarize yourself with the geographies and variables that are available for each. The API is a basic REST API, where you insert parameters into a base url and retrieve data based on the link you submit. Python has several modules you can use for interacting with APIs – the requests module is a popular choice.

The following pop estimates example is on github (but if github flops see the nbviewer example instead).

The top of the script contains basic stuff – import the modules you need, read in your key, and define the variables that you want to pull. You don’t have to use an API key, but if you don’t you’re limited to pulling in 500 records a day. Requesting a key is simple and free. A best practice is to store your key (a big integer) in a file that you read in, so you’re not exposing it in the script. Most of the census APIs require that you pass in a year and a dataset (dsource). Larger datasets may be divided into subsets (dname); for example the population estimates is divided into estimates, components of change, and characteristics (age, sex, race, etc.). Save the columns and geographies that you want to get in a comma-separated string. You have to consult the documentation and variable lists that are available for each dataset to build these, and the geography requires ANSI / FIPS codes.

%matplotlib inline
import requests,pandas as pd

with open('census_key.txt') as key:
    api_key=key.read().strip()

year='2018'
dsource='pep'
dname='components'
cols='GEONAME,NATURALINC,DOMESTICMIG,INTERNATIONALMIG'
state='42'
county='017,029,045,091,101'

Next, you can create the url. I’ve been doing this in two parts. The first part:

base_url = f'https://api.census.gov/data/{year}/{dsource}/{dname}'

Includes the base https://api.census.gov/data/ followed by parameters that you fill in. The year, data source, and dataset name are the standard pieces. The output looks like this:

'https://api.census.gov/data/2018/pep/components'

Then you take that base_url and add additional parameters that are going to vary within the script, in this case the columns and the geography, which all appear in the ‘get’ portion of the url. The ‘for’ and ‘in’ options allow you to select the type of geography within another geography, in this case counties within states, and you pass in the appropriate ANSI FIPS codes from the string you’ve created. The key appears at the end of the url, but if you opt not to use it you can omit that part. Once the link is fully constructed you use the requests module to fetch the data using that url. You can print the result out as text (assuming it’s not too long).

data_url = f'{base_url}?get={cols}&for=county:{county}&in=state:{state}&key={api_key}'
response=requests.get(data_url)
print(response.text)

The result looks like a nested list, but is actually a string that’s structured in a non-standard JSON format:

[["GEONAME","NATURALINC","DOMESTICMIG","INTERNATIONALMIG","state","county"],
["Bucks County, Pennsylvania","-178","-605","862","42","017"],
["Chester County, Pennsylvania","1829","-887","1374","42","029"],
["Delaware County, Pennsylvania","1374","-2513","1579","42","045"],
["Montgomery County, Pennsylvania","1230","-1987","2315","42","091"],
["Philadelphia County, Pennsylvania","8617","-11796","8904","42","101"]]

To do anything with it, convert it to JSON with response.json(). Then you can convert it into a list, dictionary, or in this example a Pandas dataframe. Here, I build the dataframe with everything from row one forward [1:]; row zero contains the column headers[0]. I rename some of the columns, build a unique ID by concatenating the state and county FIPS codes and set that as the new index, and drop the individual county and state FIPS columns. By default every object that’s returned is a string, so I convert the numeric columns to integers:

data=response.json()
df=pd.DataFrame(data[1:], columns=data[0]).\
    rename(columns={"NATURALINC": "Natural Increase", "DOMESTICMIG": "Net Domestic Mig", "INTERNATIONALMIG":"Net Foreign Mig"})
df['fips']=df.state+df.county
df.set_index('fips',inplace=True)
df.drop(columns=['state','county'],inplace=True)
df=df.astype(dtype={'Natural Increase':'int64','Net Domestic Mig':'int64','Net Foreign Mig':'int64'},inplace=True)
df

Then I can see the result:

pep dataframe

Once the data is in good shape, you can begin to analyze and visualize it. Here’s the components of population change for Philadelphia and the surrounding suburban counties in Pennsylvania from 2017 to 2018 – natural increase is the difference between births and deaths, and there’s net migration within the US (domestic) and between the US and other countries (foreign):

labels=df['GEONAME'].str.split(' ',expand=True)[0]
ax=df.plot.bar(rot=0, title='Components of Population Change 2017-18')
ax.set_xticklabels(labels)
ax.set_xlabel('')

Components of Population Change Plot

Each request is going to vary based on your specific needs and the construction of the particular dataset. Here’s another example where I pull data on business establishments, employees, and wages (in $1,000s of dollars) from the ZIP Code Business Patterns (ZBP). This dataset is smaller, so it doesn’t have a dataset name, just a data source. To get all the ZIP Codes in Delaware I use the asterisk * wildcard. Because ZIP Codes do not nest within states I can’t use the ‘in’ option, it’s simply not available. A state code is stored in a special field called ST, and I can use it as a general limiter with equals in the query:

year='2016'
dsource='zbp'
cols='ESTAB,EMP,PAYQTR1,PAYANN'
state='10'

base_url = f'https://api.census.gov/data/{year}/{dsource}'

data_url = f'{base_url}?get={cols}&for=zipcode:*&ST={state}&key={api_key}'
response=requests.get(data_url)
print(response.text)
[["ESTAB","EMP","PAYQTR1","PAYANN","ST","zipcode"],
["982","26841","448380","1629024","10","19713"],
["22","628","3828","15848","10","19716"],
["8","15","371","2030","10","19732"],
["7","0","0","0","10","19718"],
["738","9824","83844","353310","10","19709"]...
data=response.json()
zbp_data=pd.DataFrame(data[1:], columns=data[0]).set_index('zipcode')
zbp_data.drop(columns=['ST'],inplace=True)
for field in cols.split(','):
    zbp_data=zbp_data.astype(dtype={field:'int64'},inplace=True)
zbp_data.head()

ZBP Data for Delaware

One of the issues with the ZBP is that many variables are not disclosed due to privacy regulations; instead of returning nulls a zero is returned, but in this dataset they are not true zeros. Once you retrieve the data and set the types you can replace zeros with NaNs, which are numpy / Panda nulls – although there’s a quirk in that dataframe columns declared as integers cannot contain null values. Instead you can use a float, or a workaround that’s been implemented for new Pandas versions (for my specific use case this data will be inserted into a database, so I’ll use SQL to accomplish the zero to null conversion). ZBP data is also injected with noise to protect privacy, and you can retrieve special columns that contain noise flags.

The API is convenient for automating the data acquisition process, and allows you to cherry pick the variables you want. To avoid accessing the API over and over again as you build your scripts (which is prohibitive when requesting lots of data) you can pickle the data right after you retrieve it – a pickle is a python data object that efficiently stores data locally, and pandas has special functions for creating and accessing them. Once you pull your data and pickle it, you can comment out (or in a notebook, don’t rerun) the requests block, and subsequently pull the data from the pickle as you tweak your code (see caveat in the postscript – perhaps best to use json instead of pickle).

#Write to a pickle
zbp_data.to_pickle('insert path here.pickle')
#Read from a pickle to dataframe
zbp_new=pd.read_pickle('insert path here.pickle')

Take a look at the Census Data API User Guide to learn more. The guide focuses just on the REST API, and is not specific to a scripting language. Of course, you also need to familiarize yourself with the datasets and how they’re created and organized, and with census geography (which is why I wrote this book).

Postscript

Since I’ve finished this post I’ve created a notebook that pulls ZBP data from the API (alt nbviewer here) and have some additional thoughts I’d like to share:

  1. I decided to dump the data I retrieved from the API to a json file and then pull data from it instead of using a pickle. Pickles come with serious security issues. If you don’t intend to share your code with anyone pickles are fine, otherwise consider an alternative.
  2. My method for parsing the retrieved data into a dataframe worked fine because the census API uses non-standard JSON; essentially the string that’s returned resembles a nested Python list. If this was true JSON, we may need to employ a different method to account for the fact that the number of elements per record may vary.
  3. Wildcards are not always available to build urls for certain data; for example to download the number of establishments classified by industry I wasn’t able to grab everything for one state using the method I illustrated in this post. Instead I had to loop through a list of ZIP and NAICS codes to retrieve what I wanted one at a time.
  4. In the case of retrieving establishments classified by industry there were many cases when there was no data for a particular ZIP Code (i.e. no farms and mines in midtown Manhattan). Since I needed records that showed zero establishments, I had to insert them myself if the API returned no result. Even if you didn’t need records with zeros, it’s important to consider the potential impact of getting nothing back from the API on your subsequent code.
  5. Given my experience thus far these APIs were pretty reliable, in that I haven’t had issues with time outs and partially returned data. If this was not the case and you had lots of data to retrieve, you would need to build in some try – except statements to handle exceptions, save data as you go along, and pick up where you left off if something breaks. Read about this geocoding script I wrote a few years back for examples.
GIS consultations by status chart

Plotting Library GIS Services with Pandas

With the dawn of a new academic year I usually spend a little time looking back at the previous one. Since I began my position as Geospatial Data Librarian at Baruch College I’ve logged my questions, consultations, course visits, and workshops in a spreadsheet that I’ve used for creating summaries and charts. I spent a good chunk of this summer improving my Pandas skills, and put them to the test by summarizing and plotting my services data in a Jupyter Notebook instead.

Pandas is a data science module for Python that adds so many new components that it’s like a language all by itself. Its big selling point is that it adds a grid-like data structure to Python. In vanilla Python, you typically read data files into a list of lists where the big list represents the file, the individual lists represent rows, and the list elements represent values. There are no columns; to manipulate data you iterate through the sub-lists and elements by their position number. In well-structured datasets, elements in the same position in each sub-list represent attributes that would be stored in the same column.

In contrast, Pandas provides a true row and column structure called a dataframe, where you access each row by its index (a unique id) and columns by name or position. Furthermore, methods and functions that you apply to the data are automatically applied to entire rows and columns, and in some cases even to the entire dataframe, so that looping through data element by element is largely unnecessary. You’re able to treat a dataframe as if it were a spreadsheet or database table, in that you can concatenate dataframes together, merge them on their index numbers, and group records by values.

Using Pandas in concert with a Jupyter Notebook allows for an iterative approach to exploring and manipulating data, and is particularly conducive to creating plots and charts. You can use Python’s tried and true matplotlib module to build your chart bit by bit, or you can use Panda’s own plotting functions, which are wrappers around matplotlib that allow you to quickly create charts with fewer lines of code. Another plotting module called Seaborn offers a third approach.

This cheat sheet has become my indispensable reference for keeping track of the different Pandas functions and methods, and for helping me mentally navigate the different ways of doing things in Pandas versus regular Python. Plotting was a struggle at first, as I tried to figure out when to use Pandas versus matplotlib versus Seaborn. The fact that it’s possible to use all three at once to create the same plot added to my confusion! This visualization flowchart helped me sort things out. For simple stuff, I used the Pandas plot functions, but if the chart required additional customization I used matplotlib to generate the extra pieces, or the whole thing. In essence, use matplotlib for super detailed control over customization, and use Pandas plot functions as shortcuts for writing more concise code.

Preamble

I’ve stored my notebook and the data file on github (still a work in progress) if you’d like to take a closer look (the notebook is the ipynb file). I’m going to address a portion of what’s in the notebook in this post.

First and foremost you need to import pandas and matplotlib’s pyplot. The %matplotlib inline trick tells the notebook to display all charts that you generate with matplotlib; otherwise it just creates them without displaying them. The plt.style.use() lets you apply a global style (chart colors, background, grid lines etc) to all plots in your notebook. This convenient style sheet reference demonstrates what they all look like.

%matplotlib inline
import pandas as pd
import matplotlib.pyplot as plt
plt.style.use('seaborn-muted')

Web Stats

I’ll start with the simplest example. My spreadsheet doesn’t contain web stats, so I needed to hard code these into the notebook. To create a dataframe you build it column by column, and add the index last. In a notebook you don’t need to use a print function to see the data, you simply enter the name of the object that you want to display:

geoportal=pd.DataFrame({'Page Views' : [29500, 37254, 40421, 33527],
'Unique Views' : [23052, 29285, 31668, 26418],
'Downloads' : [3561, 6807, 6682, 5208]},
index=['2015.16','2016.17','2017.18','2018.19'])
geoportal

Dataframe in Jupyter Notebook

The plot was pretty darn simple, using Panda’s DataFrame.plot you specify the type of chart (bar in this case), pass in a few arguments, and voila! Pandas automatically uses the index for the x axis (academic years in this case) and will attempt to plot all columns on the y axis. If this isn’t desirable you can set x and y in the arguments. The default legend placement isn’t ideal in this example, but we’ll see how to change it later. plt.savefig() saves the chart as an image file outside the notebook.

geoportal.plot.bar(rot=0, title='Baruch Geoportal')
plt.savefig('webfig.png')

Baruch Geoportal Web Stats

Questions and Consultations

The rest of my data is stored in an Excel spreadsheet. You can quickly read spreadsheets into a dataframe by specifying the file and sheet, and the head() command previews the top records.

questions = pd.read_excel('RefLog.xlsx', sheet_name='Questions')
questions.head()

Dataframe of questions

I used the groupby method to summarize the number of questions by semester, indicating what column to use for grouping, and how to aggregate. In this example I use .size() which counts all records (another method called count is similar except it does not count null values). Since this result returns just a single column, Pandas returns a data type called a sequence, which is a single-column dataframe with an index (similar to a dictionary key-value pair in vanilla Python). If I want a new dataframe, I can explicitly feed in the columns, reset the index and set it to the year. You can plot data from either type.

#Summarize as a series
questions_sem=questions.groupby(by='Semester').size()

#Summarize as a dataframe
questions_yr=questions[['Year','Question']].groupby(by='Year').size().reset_index(name='Questions').set_index('Year')
questions_yr

Dataframe of questions summarized

As before, the plot is pretty simple, but in this case when saving the figure I specify bounding box tight so the labels don’t get cut off (I rotated them 45 degrees for legibility).

questions_yr.plot.bar(rot=45, legend=None, title="GIS Questions")
plt.savefig('questions.png',bbox_inches='tight')

GIS questions chart

To create a stacked bar chart that shows the number of questions and the status of the person who asked them, I can create a new dataframe where I group by both year and status. One of the initial challenges in learning how to plot data is figuring out what structure is appropriate. After some experimentation, I figured out that each status needs to be as column in order to plot it. I used the following with the unstack method to pivot the data:

 questions_status2=questions[['Year','Question','Status']]\
.groupby(by=['Year','Status']).size().unstack() questions_status2

Dataframe questions unstacked

questions_status2[['Student','Faculty','Staff','CUNY','Public']]\
.plot.bar(stacked=True, rot=45, title="GIS Questions")
plt.savefig('questions_status.png',bbox_inches='tight')

GIS questions by status chart

Explicitly stating the columns isn’t necessary, but it allows you to specify the order in which they appear in the chart. I have another worksheet that lists my consultations, that I read in, transform, and plot using the same statements:

GIS consultations by status chart

Questions represent emails or phone calls that I’ve received, while consultations are in- person, one-on-one sessions. Both the questions and consultations are specific to demographic, geospatial, or GIS-related topics. Students, faculty, and staff refer to people affiliated with my college (Baruch), while the CUNY category captures affiliates from all the other schools in the university regardless of their status. Public captures anyone outside the university.

The initial patterns are similar: the number of questions was low for my initial three years, and then began to take off in the 2010-11 academic year. This coincided with my movement out of the library’s Information Services Division and into the Graduate Services Division, where I was able to devote more time to providing my specialized services and less time providing general ones (i.e. the reference desk, visiting freshmen English composition classes). 2010-11 was also the year I introduced my day-long introductory GIS workshops which led to an increase in business, particularly from other CUNY campuses.

Another turning point was 2014-15 but the data diverges; the number of questions dips and hasn’t returned to to the peak I hit in 2013-14, while consultations remain consistently high. This is the year that I moved into the GIS Lab, and was able to provide better on-going in-person support. It was also the year I received tenure and promotion, which immediately resulted in a heavy increase in service commitments, i.e. serving on various college committees that took me away from my work (while I have graduate assistants that help with consultations, questions are sent directly to me). 2017-18 is a big divot on both charts as this was the year I was away on sabbatical to write my book (my grad assistant Janine held down the fort at the lab while I monitored questions from home), but there was a solid rebound in 2018-19.

Course Visits and Workshop Stats

I frequently visit public policy, journalism, and other courses to give lectures on census data and GIS, and for these charts I wanted to show the number of classes I visited and attendance on one chart. After loading my teaching data in, I excluded records that represented my GIS workshops by using the query method. Since I wanted to create two different aggregates – a count and a sum – I applied the .agg method after using groupby:

 classes_yr=classes[['Year','Class','Attendance']].groupby('Year').agg({'Class':'count', 'Attendance':'sum'})
classes_yr

Courses dataframe

As best as I could tell, the Pandas plot function couldn’t handle a line and bar on the same chart with a secondary Y axis, so I used matplotlib instead, building the chart one piece at a time:

plt.figure()

ax = classes_yr['Attendance'].plot(secondary_y=True, marker='o', color='orange')
ax = classes_yr['Class'].plot(kind='bar', title='Course Visits', rot=45)
ax.set_ylabel('Courses')
plt.ylabel('Attendance')

plt.savefig('courses.png',bbox_inches='tight')

Course Visits chart

The courses I visit are consistently mid-sized with about 20 students a piece, so visits and attendance track pretty closely. The pattern is similar to my questions and consultations, initially low, rising as I gained independence, dropping once I hit tenure and service commitments, then gradually rising until the 2017-18 sabbatical year.

For the GIS workshops (stored in greater detail in a separate worksheet) I wanted to create two charts: a summary of attendance for each year by status, and another showing the schools that participants came from. Since attendance will vary by the number of workshops, I also wanted to incorporate the number of sessions into the first chart. After loading in the data:

Workshops dataframeand creating a grouped summary:

Dataframe workshops summary

I created an independent sequence for the labels using string methods:

Sequence lables

and I used matplotlib so I could set different tick labels and move the legend, as the default placement blocked portions of the bars:

plt.figure()
ax=gis_yr[['Undergrad','Grad Stdt','Faculty','Staff','Other']].plot.bar\
(stacked=True, rot=25, title="GIS Workshops")

ax.set_xticklabels(gis_label)
ax.set_xlabel('Year (# Sessions)')
ax.set_ylabel('Attendance')
plt.legend(loc='upper center', bbox_to_anchor=(1, 1))

plt.savefig('workshops.png',bbox_inches='tight')

GIS workshops chart

For the workshops, the status includes all CUNY members regardless of school, while Other is anyone not affiliated with CUNY. Graduate students have always comprised the largest share of participants. Once again, there is the tenure dip in 2014-15 (fewer sessions) and no sessions during 2017-18 sabbatical. 2016-17 was an exceptional year as one of our sessions was held at the FOSS4G conference, so there are lots of participants from the Other category. The latest year was disappointing, as bad weather impacted attendance at two of the sessions.

I wanted to create a pie chart to show participation by CUNY school, but to make it aesthetically pleasing I needed to remove schools with few participants and add them to an Other CUNY category. Otherwise there would be tiny wedges with unreadable labels. After creating a subset of the workshops dataframe that summed values only for the school columns, I iterated through the schools to sum attendance to a variable, dropped those schools, and added the sum to the other category (see the notebook for details). I used the Pandas plot function to create the pie chart, and used the autopct argument to display percentages in the wedges. I also specified a figure size, which you can do for any chart (and becomes important when you decide to embed them in documents):

gis_total=gis_schools.sum()

gis_schools.plot.pie(legend=False, figsize=(6,6), \
title='Workshop Participants by School \n ({} Participants in Total)'.format(gis_total), autopct='%i%%')
plt.ylabel("")
plt.savefig('schools.png',bbox_inches='tight')

Pie chart showing workshop participation

One-third of participants were from my college, and one-fourth were from the Graduate Center, which is our nearest CUNY neighbor with a large population of master’s and PhD students who are keenly interested in learning GIS. The next biggest contributors are Hunter and Lehman Colleges, which are the two CUNY schools that have geography departments with GIS programs; Hunter is also close to Baruch, and we took a road trip to offer some sessions on Lehman’s campus.

Wrap Up

What I like about this approach is that you can summarize and reconfigure data without messing with the original source, and you can clearly see what your formulas are as they’re not hidden beneath the resulting values. These are both hazards when working directly within spreadsheets. While it takes time to learn these new functions and to grapple with finding work-arounds for exceptions, I don’t think it’s any less difficult than trying to accomplish the same things in a spreadsheet. I’ve always found spreadsheet charting to be rather clumsy, where you’re forced to cycle through numerous windows or to click on minuscule pieces of a chart to access hidden settings that you need.  The Pandas / notebook approach makes a lot of sense for iterative data exploration, summation, and visualization, although I’ll continue to rely on regular Python for projects that fall outside this specific domain.

Net Out-Migration from the NY Metro Area to Other Metro Areas 2011-2015

Recent Migration Trends for New York City and Metro

The Baruch GIS lab crew just published a paper: New Yorkers on the Move: Recent Migration Trends for the City and Metro Area. The paper (no. 15 Feb 2018) is part of the Weissman Center for International Business Occasional Paper Series, which focuses on New York City’s role in the international and domestic economy.

Findings

We analyzed recent population trends (2010 to 2016) in New York City and the greater metropolitan area using the US Census Bureau’s Population Estimates to study components of population change (births, deaths, domestic and international migration) and the IRS Statistics of Income division’s county to county migration data to study domestic migration flows.

Here are the main findings:

  1. The population of New York City and the New York Metropolitan Area increased significantly between 2010 and 2016, but annually growth has slowed due to greater domestic out-migration.
  2. Compared to other large US cities and metro areas, New York’s population growth depends heavily on foreign immigration and natural increase (the difference between births and deaths) to offset losses from domestic out-migration.
  3. Between 2011 and 2015 the city had few relationships where it was a net receiver of migrants (receiving more migrants than it sends) from other large counties. The New York metro area had no net-receiver relationships with any major metropolitan area.
  4. The city was a net sender (sending more migrants than it received) to all of its surrounding suburban counties and to a number of large urban counties across the US. The metro area was a net sender to metropolitan areas throughout the country.

For the domestic migration portion of the analysis we were interested in seeing the net flows between places. For example, the NYC metro area sends migrants to and receives migrants from the Miami metro. What is the net balance between the two – who receives more versus who sends more?

The answer is: the NYC metro is a net sender to most of the major metropolitan areas in the country, and has no significant net receiver relationships with any other major metropolitan area. For example, for the period from 2011 to 2015 the NYC metro’s largest net sender relationship was with the Miami metro. About 88,000 people left the NYC metro for metro Miami while 58,000 people moved in the opposite direction, resulting in a net gain of 30,000 people for Miami (or in other words, a net loss of 30k people for NYC). The chart below shows the top twenty metros where the NYC metro had a deficit in migration (sending more migrants to these areas than it received). A map of net out-migration from the NYC metro to other metros appears at the top of this post. In contrast, NYC’s largest net receiver relationship (where the NYC metro received more migrants than it sent) was with Ithaca, New York, which lost a mere 300 people to the NYC metro.

All of our summary data is available here.

domestic migration to NYMA 2011-2015: top 20 deficit metro areas

Process

For the IRS data we used the county to county migration SQLite database that Janine meticulously constructed over the course of the last year, which is freely available on the Baruch Geoportal. Anastasia employed her Python and Pandas wizardry to create Jupyter notebooks that we used for doing our analysis and generating our charts, all of which are available on github. I used an alternate approach with Python and the SQLite and prettytable modules to generate estimates independently of Anastasia, so we could compare the two and verify our numbers (we were aggregating migration flows across years and geographies from several tables, and calculating net flows between places).

One of our goals for this project was to use modern tools and avoid the clunky use of email. With the Jupyter notebooks, git and github for storing and syncing our work, and ShareLaTeX for writing the paper, we avoided using email for constantly exchanging revised versions of scripts and papers. Ultimately I had to use latex2rtf to convert the paper to a word processing format that the publisher could use. This post helped me figure out which bibliography packages to choose (in order for latex2rtf to interpret citations and references, you need to use the older natbib & bibtex combo and not biblatex & biber).

If you are doing similar research, Zillow has an excellent post that dicusses the merits of the different datasets. There are also good case studies on Washington DC and Philadelphia that employ the same datasets.

Python Geocoding Take 2 – US Addresses

Python Geocoding Take 1 – International Addresses I discussed my recent adventures with geocoding addresses outside the US. In contrast, there are countless options for batch geocoding addresses within the United States. I’ll discuss a few of those options here, but will focus primarily on the US Census Geocoder and a Python script I’ve written to batch match addresses using their API. The code and documentation is available on my lab’s resources page.

A Few Different Options

ESRI’s geocoding services allow you (with an account) to access their geocoding servers through tools in the ArcToolbox, or you can write a script and access them through an API. QGIS has a third-party plugin for accessing Google’s services (2500 records a day free) or the Open Streetmap. You can still do things the old fashioned way, by downloading geocoded street files and creating a matching service.

Alternatively, you can subscribe to any number of commercial or academic services where you can upload a file, do the matching, and download results. For years I’ve used the geocoding services at Texas A&M that allow you to do just that. Their rates are reasonable, or if you’re an academic institution and partner with them (place some links to their service on their website) you can request free credits for doing matches in batches.

The Census Geocoder and API, and a Python Script for Batch Geocoding

The Census Bureau’s TIGER and address files are often used as the foundational layers for building these other services, to which the service providers add refinements and improvements. You can access the Census Bureau’s services directly through the Census Geocoder, where you can match an address one at a time, or you can upload a batch of 1000 records. It returns longitude and latitude coordinates in NAD 83, and you can get names and codes for all the census geographies where the address is located. The service is pretty picky about the structure of the upload file (must be plain text, csv, with an id column and then columns with the address components in a specific order – with no other attributes allowed) but the nice thing is it requires no login and no key. It’s also public domain, so you can do whatever you want with the data you’ve retrieved. A tutorial for using it is available on our lab’s census tutorials page.

census geocoder

They also have an API with some basic documentation. You can match parsed and unparsed addresses, and can even do reverse geocoding. So I took a stab at writing a script to batch process addresses in text-delimited files (csv or txt). Unfortnately, the Census Geocoding API is not one of the services covered by the Python Geocoder that I mentioned in my previous post, but I did find another third party module called censusgeocode which provides a thin wrapper you can use. I incorporated that module into my Python 3 script, which I wrote as a function that takes the following inputs:

census_geocode(datafile,delim,header,start,addcol)
(str,str,str,int,list[int]) -> files

  • datafile – this is the name of the file you want to process (file name and extension). If you place the geocode_census_funct.py file in the same directory as your data file, then you just need to provide the name of the file. Otherwise, you need to provide the full path to the file.
  • delim – this is the delimiter or character that separates the values in your data file. Common delimiters includes commas ‘,’, tabs ‘t’, and pipes ‘|’.
  • header – here you specify whether your file has a header row, i.e. column names. Enter ‘y’ or ‘yes’ if it does, ‘n’ or ‘no’ if it doesn’t.
  • start – type 0 to specify that you want to start reading the file from the beginning. If you were previously running the script and it broke and exited for some reason, it provides an index number where it stopped reading; if that’s the case you can provide that index number here, to pick up where you left off.
  • addcol – provide a list that indicates the position number of the columns that contain the address components in your data file. For an unparsed address, you provide just one position number. For a parsed address, you provide 4 positions: address, city, state, and ZIP code. Whether you provide 1 or 4, the numbers must be supplied in brackets, as the function requires a Python list.

You can open the script in IDLE, run it to load it into memory, and then type the function with the necessary parameters in the shell to execute it. Some examples:

  • A tab-delimited, unparsed address file with a header that’s stored in the same folder as the script. Start from the beginning and the address is in the 2nd column: census_geocode('my_addresses.txt','t','y',0,[2])
  • A comma-delimited, parsed address file with no header that’s stored in the same folder as the script. Start from the beginning and the addresses are in the 2nd through 5th columns: census_geocode('addresses_to_match.csv',',','n',0,[2,3,4,5])
  • A comma-delimited, unparsed address file with a header that’s not in the same folder as the script. We ran the file before and it stopped at index 250, so restart there – the address is in the 3rd column: census_geocode('C:address_datadata1.csv',',','y',250,[3])

The beginning of the script “sets the table”: we read the address columns into variables, create the output files (one for matches, one for non-matches, and a summary report), and we handle whether or not there’s a header row. For reading the file I used Python’s CSV module. Typically I don’t use this module, as I find it’s much simpler to do the basic: read a line in, split it on a delimiter, strip whitespace, read it into a list, etc. But in this case the CSV module allows you to handle a wider array of input files; if the input data was a csv and there happened to be commas embedded in the values themselves, the CSV module easily takes care of it; if you ignore it, the parsing would get thrown off for that record.

Handling Exceptions and Server Errors

In terms of expanding my skills, the new things I had to learn were exception handling and control flows. Since the censusgeocoding module is a thin wrapper, it had no built in mechanism for retrying a match a certain number of times if the server timed out. This is an absolute necessity, because the census server often times out, is busy, or just hiccups, returning a generic error message. I had already learned how to handle crashes in my earlier geocoding experiments, where I would write the script to match and write a record one by one as it went along. It would try to do a match, but if any error was raised, it would exit that loop cleanly, write a report, and all would be saved and you could pick up where you left off. But in this case, if that server non-response error was returned I didn’t want to give up – I wanted to keep trying.

So on the outside there is a loop to try and do a match, unless any error happens, then exit the loop cleanly and wrap up. But inside there is another try loop, where we try to do a match but if we get that specific server error, continue: go back to the top of that for loop and try again. That loop begins with While True – if we successfully get to the end, then we start with the next record. If we get that server error we stay in that While loop and keep trying until we get a match, or we run out of tries (5) and write as a non-match.

error handling

In doing an actual match, the script does a parsed or unparsed match based on user input. But there was another sticking point; in some instances the API would return a matched result (we got coordinates!), but some of the objects that it returned were actually errors because of some java problem (failed to get the tract number or county name – here’s an error message instead!) To handle this, we have a for i in range loop. If we have a matched record and we don’t have a status message (that indicates an error) then we move along and grab all the info we need – the coordinates, and all the census geography where that coordinate falls, and write it out, and then that for loop ends with a break. But if we receive an error message we continue – go back to the top of that loop and try doing the match again. After 3 tries we give up and write no match.

Figuring all that out took a while – where do these loops go and what goes in them, how do I make sure that I retry a record rather than passing over it to the next one, etc. Stack Exchange to the rescue! Difference between continue, pass and break, returning to the beginning of a loop, breaking out of a nested loop, and retrying after an exception. The rest is pretty straightforward. Once the matching is done we close the files, and write out a little report that tells us how many matches we got versus fails. The Census Geocoder via the API is pretty unforgiving; it either finds a match, or it doesn’t. There is no match score or partial matching, and it doesn’t give you a ZIP Code or municipal centroid if it can’t find the address. It’s all or nothing; if you have partial or messy addresses or PO Boxes, it’s pretty much guaranteed that you won’t get matches.

There’s no limit on number of matches, but I’ve built in a number of pauses so I’m not hammering the server too hard – one second after each match, 5 seconds after every 1000 matches, a couple seconds before retrying after an error. Your mileage will vary, but the other day I did about 2500 matches in just under 2 hours. Their server can be balky at times – in some cases I’ve encountered only a couple problems for every 100 records, but on other occasions there were hang-ups on every other record. For diagnostic purposes the script prints every 100th record to the screen, as well as any problems it encountered (see pic below). If you launch a process and notice the server is hanging on every other record and repeatedly failing to get matches, it’s probably best to bail out and come back later. Recently, I’ve noticed fewer problems during off-peak times: evenings and weekends.

script_running

Wrap Up

The script and the documentation are posted on our labs resources page, for all to see and use – you just have to install the third party censusgeocode module before using it. When would you want to use this? Well, if you need something that’s free, this is a good choice. If you have batches in the 10ks to do, this would be a good solution. If you’re in the 100ks, it could be a feasible solution – one of my colleagues has confirmed that he’s used the script to match about 40k addresses, so the service is up to the task for doing larger jobs.

If you have less than a couple thousand records, you might as well use their website and upload files directly. If you’re pushing a million or more – well, you’ll probably want to set up something locally. PostGIS has a TIGER module that lets you do desktop matching if you need to go into the millions, or you simply have a lot to do on a consistent basis. The excellent book PostGIS in Action has a chapter dedicated to to this.

In some cases, large cities or counties may offer their own geocoding services, and if you know you’re just going to be doing matches for your local area those sources will probably have greater accuracy, if they’re adding value with local knowledge. For example, my results with NYC’s geocoding API for addresses in the five boroughs are better than the Census Bureau’s and is customized for local quirks; for example, I can pass in a borough name instead of a postal city and ZIP Code, and it’s able to handle those funky addresses in Queens that have dashes and similar names for multiple streets (35th st, 35th ave, 35th dr…). But for a free, public domain service that requires no registration, no keys, covers the entire country, and is the foundation for just about every US geocoding platform out there, the Census Geocoder is hard to beat.

Python Geocoding Take 1 – International Addresses

This past semester has been the semester of geocoding. I’ve had a number of requests for processing large batches of addresses. Now that the term is drawing to a close, I’ll share some of my trials and tribulations. In this post, I’ll focus on my adventures in international geocoding.

First, it’s necessary to provide some context. As an academic librarian I’m primarily engaged with assisting students and faculty with their coursework and their research. My users are interested in getting coordinates for data so that they can do both analysis and visualization, which requires them to download the actual coordinate data in a batch and integrate it with the rest of their projects.

This is an important distinction to make, because in many cases the large web mapping companies (Google, Bing, Mapquest, etc) are not catering to this population – they provide services and APIs to web developers, so these folks can integrate geocoding services into the Google, Bing, etc maps they are embedding in their website. They geocoding providers specifically prohibit (in the fine print of their terms of use) anyone from using their services to create and download geocoded data. This essentially excludes a lot of academic use – which, is something I hadn’t fully grasped at the outset.

Google’s Geocoding API Perhaps?

My adventure began when a professor asked me for help in geocoding about 1 million addresses – in Turkey. Right from the beginning, many of the usual sources I would turn to (for US addresses) were out the window. I knew that I could do small scale batches of international addresses with the mmQGIS geocoding plugin, so I started testing there. The address file we had consisted of unparsed addresses, and the formating looked rather chaotic – but after doing some research I discovered that geocoding Turkish addresses was a tough proposition. The Open Street Map plugin (using Nominatim) returned no matches for our 1000 test cases. The Google results were much better, so we decided to investigate writing a script and using an API and to pay for the matching. According to the documentation, it would end up costing $500 to do 1 million addresses.

I searched around for some Python APIs and found what I believed was the official one for google maps geocoding. So I spent a day writing a script that would loop through the addresses, which we divided into batches of 100k records each (which is the max you can do per day with Google if you set up billing), and the professor obtained an API key and set up billing for the account. The interface for setting up and managing the Google APIs was ridiculously confusing. Eventually we were set and I let the script rip, and found that it wouldn’t rip for long. It would consistently stop after doing a few thousand records. I had written it to write results one by one as they were obtained, and to exit cleanly in the case of errors. Upon exit, it provides the index number of the record where it stopped, so I was able to pick up where it left off. But the server would constantly time out – sometimes it could do 10 to 12k records in a stretch, but often less, so I could never leave it unattended for long. The matches themselves were a mixed bag – you could throw absolute garbage at the Google geocoder and still get a match – if not to an address or property, then to a street segment, and beyond that to useless things like postal codes, administrative districts, and the country as a whole (i.e. I can’t find your address, so here are the coordinates for the geographic center of Istanbul, or for all of Turkey. Have a nice day).

It seemed like it was going to be a long climb to get to 1 million – but after about 100k we could go no further. Google simply refused every additional request. A new API key would get us a little further, but soon after that nothing would work and we wouldn’t get any useful error messages to explain why. Having never done anything like this before, I started to investigate why, and eventually discovered the problem: these web mapping geocoding services, even if you pay for them, are not meant to be used this way. Buried in the documentation I found the license restrictions, which stipulate that you are not allowed to download any of the data, and you had to plot every coordinate you retrieved onto a Google map. This is a service for web mapping developers, not researchers.

Why hadn’t I realized this before? One, I simply had never made this distinction as I thought geocoding was geocoding, and in my world of course people are going to want to download the coordinates. Two, the Internet is full of thousands of little blog posts and tutorials which demonstrate how to use the Google Maps APIs, so I thought this was possible. But they never mention any of the caveats about what you can and can’t do with these services. In addition to violating the service terms, what I was doing was akin to yelling in the back of a crowded room, as I was hammering their server, sending requests as fast as I could with no limit. A normal web mapping application (which is what the service is designed for) would send a fraction of those requests in that amount of time. No wonder the requests were refused. Thus ended my Google geocoding experiment.

Nope – How About ESRI Instead?

So what to do next? I found that most of the other commercial web mapping services didn’t provide anything near the maximum caps and low prices that Google was offering. Mapquest for example requires that you subscribe to an account on a monthly or annual basis, and 100k is the amount you could do in a month. Most of the other commercial services also prohibit any downloads.

The big exception is ESRI – they are one of the few that understand and cater to the academic market, and they do allow downloads: they say quite plainly: “Take your Coordinates with you. Once you have the results of a Geocode operation, they’re yours to take anywhere.” My university has a site license for ArcGIS, but it doesn’t include geocoding. You can create an account and have a certain number of free credits, and after that you pay. 1 mil records was going to cost about $4000 – substantially more than Google, but totally legal. ESRI provides lists of countries and ranks them according to how complete their street network coverage is. You can use their API via a script, or you can set up the service in ArcGIS Desktop and do the matching through the ArcToolbox. This would be painfully slow if you were doing a large job (like this one) but for the purpose of testing it out with a few hundred records this is what I tried. Unfortunately, in our case the results still weren’t good. Most of the addresses were to administrative or postal areas; not specific enough.

The Python Geocoder and a Wealth of Options

What often happens in librarianship when a patron makes an initial request (this should be a piece of cake, right?) and then discovers that what they’re looking for is more involved (ahhh this will be tougher than we thought), is that they reframe the question. He went back through the addresses with a research partner and winnowed them down based on what they really, absolutely needed, so now we were down from 1 million to just finding a match for about 300k. His colleague also suggested that we use Yandex, the Russian search and mapping engine. The structure of Russian addresses is quite similar to Turkish ones, and since Russia is closer to Turkey geographically and economically Yandex might do a better job.

I was dubious of this at first, but was quickly surprised. I found the Python Geocoder module, which provides a common, uniform API to over two dozen different geocoding services – including Google. Given the simplicity and flexibility of this module, it’s the one I should have used in the first place. And while Google limits you to 2500 free matches in one day, Yandex allows you 25k – that’s 25,000 – free matches in one day, without having to request an API key! I modified the original script I wrote to use the Python Geocoder module with Yandex, and the initial small-batch tests were successful. Here’s a small portion of the code – it loops through a file where the address is stored in one field (unparsed):

for index, line in enumerate(readfile):
        address=line.strip().split(delim)
        result=geocoder.yandex(address[add]).json

And it spits you back this JSON result (you could also do XML if you prefer):

{‘quality’: ‘street’, ‘address’: ‘Türkiye, İstanbul, Fatih, Cankurtaran Mh., Ayasofya Meydanı’, ‘location’: ‘Hagia Sophia Museum, Sultanahmet Mh., Ayasofya Meydanı, Fatih/İstanbul’, ‘state’: ‘İstanbul’, ‘lng’: ‘28.979031’, ‘accuracy’: ‘street’, ‘encoding’: ‘utf-8’, ‘provider’: ‘yandex’, ‘country_code’: ‘TR’, ‘ok’: True, ‘status_code’: 200, ‘lat’: ‘41.00772’, ‘country’: ‘Türkiye’, ‘county’: ‘Fatih’, ‘confidence’: 10, ‘bbox’: {‘northeast’: [41.008156, 28.979714], ‘southwest’: [41.007285, 28.978349]}, ‘street’: ‘Ayasofya Meydanı’, ‘status’: ‘OK’}

If the result you get back is not OK (ok is False – nothing matched), then write the record to the unmatched file. Otherwise, get the bits and pieces out of the json object that you want, append them to the record, and write the whole record out to a matched file.

        if result.get('ok')==False:
                nomatch.append(address)
                nomatchfile.writelines('t'.join(address)+'n')
        else:
                lng=result.get('lng')
                lat=result.get('lat')
                qual=result.get('quality')
                accu=result.get('accuracy')
                matchadd=result.get('address')
                newitems=lng,lat,qual,accu,matchadd
                address.extend(newitems)
                matched.append(address)
                matchfile.writelines('t'.join(address)+'n')

But is it legal? It was unclear to me; they specify that map data is meant for personal/noncommercial use and in the same sentence: “Any copying of the Data, their reproduction, conversion, distribution, promulgation (publication) in the Internet, any use of the Data in mass media and/or for commercial purposes without a prior written consent of the right holder, shall be prohibited”. Does that mean any copying, or just copying for commercial use or for redistributing the data? In our case, this is for academic non-profit use and the data (individual geocoded records) wasn’t going to be republished – it would be used for plotting distances between locations and making highly generalized static dot maps for an article. At this stage we seemed to be out of options – if you need to geocode a large batch of international addresses, AND you are willing to pay for it, where on Earth can you go?

Ultimately, I left it up to the professor to contact them or not, and we decided to roll the dice. For my part, I engineered the script to put a minimum load on their servers – essentially I could take 24 hours to do 25k records. I used the time and random modules in Python to build pauses in between records to slow things down. In sharp contrast to Google, the Yandex servers were amazingly reliable – they were able to do batches of 25k records every single time without timing out – not even once – and in less than a couple weeks we were finished. About 50% of the matches were good, and for the others he and a research assistant went back and cleaned up unmatched records, and I gave them the script so they could try again.

International Geocoding: The Take-Aways

  1. If you need to geocode a large batch of foreign addresses for academic or research purposes, forget Google. Their service was less than stellar (to put it mildly) and anyway it’s a violation of their license agreement. And all those lousy little blog posts out there that show you how to use the Google Map APIs with Python and say “Gee isn’t this great!” are largely useless for practical purposes.
  2. The Python Geocoder module is simple to use and let’s you write a single script to access a ton of different geocoding services, including Open Streetmap, Yandex, and ESRI. But you still need to review the terms of service for each one to see what’s allowed and what the daily limits are.
  3. If you have funding for your research project, and ESRI geocoding has good coverage for your geographic area (based on their documentation but also on your own testing) then go with them, as you’re free and clear to download data under their terms. Arc Desktop will be too sluggish for large batches so write a script – you can use the Python Geocoder.
  4. Otherwise – the Open Street Map / Nominatim services are worth a try but your success will vary by country. I had used them before for addresses in France with fair success, but it didn’t help me with Turkey.
  5. You can also crawl through the GIS Stackexchange for advice. I’ve found that most of the suggestions are either for US geocoding, or are companies that are answering posts saying “Hey you can try my service!”

Happy geocoding, comrades! In my next post I’ll discuss my experience with batch geocoding addresses here in the US of A with Python.

Writing Functions and Building a Jinja Template

In previous posts I demonstrated how to pull data from a sqlite / spatialite data to generate reports using Python and Jinja, where Jinja2 is used as a template engine for creating LaTeX documents and the NYC Geodatabase is used as my test case. Up until now the scripts pulled the data “as is”. In this post I’ll demonstrate how I created derived variables, and how I created the Jinja2 template for the report. Please note – instead of duplicating all of the code I’m just going to illustrate the new pieces – you should check out the earlier posts to see how all the pieces fit together.

Aggregating Variables

Aggregating census data is a pretty common operation, and when working with American Community Survey estimates it’s also necessary to calculate a new margin of error for each derived value. I wrote two functions to accomplish this. For each function you pass in the keys for values you want to aggregate, a name which will be the name of the new variable, and a dictionary that contains all the keys and values that were taken from a database table for a specific geography.

#Functions for summing individual values and calculating margins of error
#for individual values

def calc_sums(keys,name,adict):
    tosum=[]
    for val in keys:
        tosum.append(adict.get(val))
    agg=sum(tosum)
    adict[name]=agg

def calc_moe(keys,name,adict):
    sqrd=[]
    for val in keys:
        item=adict.get(val)
        if item=='':
            pass
        else:
            sqrd.append(item**2)
    moe=round(math.sqrt(sum(sqrd)))
    adict[name]=moe

Later in the script, as we’re looping through all the geographies and gathering the necessary data into dictionaries that represent each data table, we call the function. In this example we’re combining household income brackets so that we don’t have so many categories:

for geog in geodict.keys():

    name=geodict.get(geog)
    filename='zzpuma_' + geog + '.tex'
    folder='puma_rept'
    outpath=os.path.join(folder,filename)

    acs1dict=pulltab('b_pumas_2013acs1','GEOID2',geog)
    acs2dict=pulltab('b_pumas_2013acs2','GEOID2',geog)

    calc_sums(['INC03_E','INC04_E'],'INC10K_E',acs1dict)
    calc_moe(['INC03_M','INC04_M'],'INC10K_M',acs1dict)
    calc_sums(['INC05_E','INC06_E'],'INC25K_E',acs1dict)
    calc_moe(['INC05_M','INC06_M'],'INC25K_M',acs1dict)
    calc_sums(['INC09_E','INC10_E'],'INC100K_E',acs1dict)
    calc_moe(['INC09_M','INC10_M'],'INC100K_M',acs1dict)

Rather than creating a new dictionary, these new values are simply appended to the existing dictionaries that contain the data taken from each of the ACS data tables in the database. They can be referenced in the template using their new column name.

Calculating Areas

I also want to include the geographic size of the PUMA as one of the report items. Columns for the area are included in the spatial table for the PUMAs – the features originally came from the TIGER files, and all TIGER files have an ALAND and an AWATER column that has land and water area in square meters. So we don’t have to calculate the area from the geometry – we can just use this function to convert the land and water attributes to square miles, and then calculate a total area:

def calc_area(adict,land,water,total):
    landarea=round(adict.get('aland10')*0.000000386102,2)
    waterarea=round(adict.get('awater10')*0.000000386102,2)
    totalarea=landarea+waterarea
    adict[land]=landarea
    adict[water]=waterarea
    adict[total]=totalarea

In the body of our script, we invoke our pulltab function (explained in an earlier post) to grab all the data from the PUMA spatial boundary table:

area=pulltab('c_bndy_pumas2010','geoid10',geog)

And then we can call our area function. We pass in the area dictionary, and what we want the new output column names to be – area for land, water, and total:

calc_area(area, 'LAND_SQM','WAT_SQM','TOT_SQM')

Like our previous aggregate script, this function appends our new values to the existing table-dictionary – in this case, one called area.

Aggregating Geographies

Our last function is a little more complicated. In all of our previous examples, we pulled PUMA-level data from the American Community Survey tables. What if we wanted 2010 Census data for the PUMAs? Decennial census data is not tabulated at the PUMA level, but it is tabulated at the census tract level. Since PUMAs are created by aggregating tracts, we can aggregate the census tract data in the NYC Geodatabase into PUMAs. Here’s our function:

#Function aggregates all values in a table with a group by field from a
#joined table, then creates a dictionary consisting of column names and values
#for a specific geography

def sumtab(tabname,jointab,id1,id2,gid,geog):
    query='SELECT * FROM %s LIMIT 1' %(tabname)
    curs.execute(query)
    col_names = [cn[0] for cn in curs.description]
    tosum=[]
    for var in col_names[3:]:
        tosum.append("SUM("+var+") AS '0_"+var+"'")
    summer=', '.join(str(command) for command in tosum)
    query='SELECT %s, %s FROM %s, %s WHERE %s = %s and %s = %s GROUP BY %s' %(gid,summer,tabname,jointab,id1,id2,gid,geog,gid)
    curs.execute(query)
    col_names = [cn[0] for cn in curs.description]
    rows = curs.fetchall()
    for row in rows:
        thedict=dict(zip(col_names,row))
    return thedict

What’s going on here? The first thing we need to do is associate the census tracts with the PUMAs they’re located in. The NYC Geodatabase does NOT have a relationship table for this, so I had to create one. We have to pass in the table name, the relationship table, the unique IDs for each, and then the ID and the geography that we’re interested in (remember our script is looping through PUMA geographies one by one). The first thing we do is a little trick – we get the names of every column in the existing data table, and we append them to a list where we create a new column name based on the existing one (in this case, append a 0 in front of the column name – in retrospect I realize this is a bad idea as column names should not begin with numbers, so this is something I will change). Then we can take the list of column names and create a giant string out of them.

With that giant string (called summer) we can now pass all of the parameters that we need into the SQL query. This selects all of our columns (using the summer string), the table names and join info, for the specific geographic area that we want and then groups the data by that geography (i.e. all tracts that have the same PUMA number). Then we zip the column names and values together in a dictionary that the function returns.

Later on in our script, we call the function:

   census10=sumtab('b_tracts_2010census','b_tracts_to_pumas','GEOID2','tractid','pumaid',geog)

Which creates a new dictionary called census10 that has all the 2010 census data for our PUMA. Like the rest of our dictionaries, census10 is passed out to the Jinja2 template and its values can be invoked using the dictionary keys (the column headings):

outfile=open(outpath,'w')
    outfile.write(template.render(geoid=geog, geoname=name, acs1=acs1dict, acs2=acs2dict, area=area,
                                  c2010=census10))
    outfile.close()

Designing the Template

The Jinja template is going to look pretty busy compared to our earlier examples, and in both cases they’re not complete (this is still a work in progress).

I wanted to design the entire report first, to get a sense for how to balance everything I want on the page, without including any Jinja code to reference specific variables in the database. So I initially worked just in LaTeX and focused on designing the document with placeholders. Ultimately I decided to use the LaTeX minipage environment as it seemed the best approach in giving me control in balancing items on the page. The LaTeX wikibook entries on floats, figures, and captions and on boxes was invaluable for figuring this out. I used rule to draw boxes to serve as placeholders for charts and figures. Since the report is being designed as a document (ANSI A 8 1/2 by 11 inches) I had no hang-up with specifying precise dimensions (i.e. this isn’t going into a webpage that could be stretched or mushed on any number of screens). I loaded the xcolor package so I could modify the row colors of the tables, as well as a number of other packages that make it easy to balance table and figure captions on the page (caption, subscaption, and multicol).

Once I was satisfied with the look and feel, I made a copy of this template and started modifying the copy with the Jinja references. The references look awfully busy, but this is the same thing I’ve illustrated in earlier posts. We’re just getting the values from the dictionaries we created by invoking their keys, regardless of whether we’re taking new derived values that we created or simply pulling existing values that were in the original data tables. Here’s a snippet of the LaTeX with Jinja that includes both derived (2010 Census, area) and existing (ACS) variables:

%Orientation - detail map and basic background info
begin{minipage}{textwidth}
	begin{minipage}[h]{3in}
   		centering
   		rule{3in}{3in}
    		captionof{figure}{Race by 2010 Census Tract}
	end{minipage}
  	hfill
	begin{minipage}[h]{4in}
		centering
		captionof{table}{Geography}
    	begin{tabular}{cccc}hline
		& Land & Water & Total\ hline
		Area (sq miles) &  num{VAR{area.get('LAND_SQM')}} & num {VAR{area.get('WAT_SQM')}} &  num {VAR{area.get('TOT_SQM')}} \ hline
		vspace{10pt}
	 end{tabular}
	captionof{table}{Basic Demographics}
	rowcolors{3}{SpringGreen}{white}
	 begin{tabular}{cccc}hline
		& textbf{2010 Census} & textbf{2009-2013} & textbf{ACS Margin}\
		& & textbf{ACS} & textbf{of Error}\ hline
		Population & num {VAR{c2010.get('0_HD01_S001')}} & num{VAR{acs2.get('SXAG01_E')}} & +/- num{VAR{acs2.get('SXAG01_M')}}\
		Males & num {VAR{c2010.get('0_HD01_S026')}} & num{VAR{acs2.get('SXAG02_E')}} & +/- num{VAR{acs2.get('SXAG02_M')}}\
		Females & num {VAR{c2010.get('0_HD01_S051')}}& num{VAR{acs2.get('SXAG03_E')}} & +/- num{VAR{acs2.get('SXAG03_M')}}\
		Median Age (yrs) & num {99999} & num{VAR{acs2.get('SXAG17_E')}} & +/- num{VAR{acs2.get('SXAG17_M')}}\
		Households & num {VAR{c2010.get('0_HD01_S150')}} & num{VAR{acs1.get('HSHD01_E')}} & +/- num{VAR{acs1.get('HSHD01_M')}}\
		Housing Units & num {VAR{c2010.get('0_HD01_S169')}} & num{VAR{acs2.get('HOC01_E')}} & +/- num{VAR{acs2.get('HOC01_M')}}\ hline
	end{tabular}
	end{minipage}
 end{minipage}

And here’s a snippet of the resulting PDF:

report_inprogress

What Next?

You may have noticed references to figures and charts in some of the code above. I’ll discuss my trials and tribulations with trying to use matplotlib to create charts in some future post. Ultimately I decided not to take that approach, and was experimenting with using various LaTeX packages to produce charts instead.