Make single CLI tool be an entry point to the taxcalc conda package #1253

martinholmer · 2017-03-20T22:16:27Z

This pull request changes the approach to developing a command-line interface (CLI) to Tax-Calculator to be in line with several suggestions @MattHJensen has offered over the past several weeks. Thanks, Matt.

Now the CLI, which is called tc, is an integral part of the conda taxcalc package, which makes it easy to install after a user has install Anaconda on his/her computer:

conda install taxcalc
conda install bokeh         <---- needed as long as bokeh is not a dependency of taxcalc

After that installation, the command conda list taxcalc shows the Tax-Calculator version number and the command tc --help produces this:

usage: tc INPUT TAXYEAR [--reform REFORM] [--assump  ASSUMP]
                        [--exact] [--graph] [--ceeu] [--dump]

Writes to a file the federal income and payroll tax OUTPUT for each filing
unit specified in the INPUT file, with the OUTPUT computed from the INPUT for
the TAXYEAR using Tax-Calculator. The OUTPUT file is a CSV-formatted file that
contains tax information for each INPUT filing unit.

positional arguments:
  INPUT            INPUT is name of CSV-formatted file that contains for each
                   filing unit variables used to compute taxes for TAXYEAR.
  TAXYEAR          TAXYEAR is calendar year for which taxes are computed.

optional arguments:
  -h, --help       show this help message and exit
  --reform REFORM  REFORM is name of optional JSON reform file. No --reform
                   implies use of current-law policy.
  --assump ASSUMP  ASSUMP is name of optional JSON economic assumptions file.
                   No --assump implies use of no customized assumptions.
  --exact          optional flag that suppresses the smoothing of "stair-step"
                   provisions in the tax law that complicate marginal-tax-rate
                   calculations.
  --graph          optional flag that causes graphs to be written to HTML
                   files for viewing in browser.
  --ceeu           optional flag that causes normative welfare statistics,
                   including certainty-equivalent expected-utility (ceeu) of
                   after-tax income values for different constant-relative-
                   risk-aversion parameter values, to be written to screen.
  --dump           optional flag that causes OUTPUT to contain all INPUT
                   variables (possibly extrapolated to TAXYEAR) and all
                   calculated tax variables, where all the variables are named
                   using their internal Tax-Calculator names. No --dump option
                   implies OUTPUT contains minimal tax output.

This pull request also adds a file-upload-tests/cli-results bash script that uses the tc CLI to produce results that can be compared to results generated by the TaxBrain file-upload capability as explained in the file-upload-tests/README.md file in this pull request.

@MattHJensen @feenberg @Amy-Xu @andersonfrailey @GoFroggyRun @codykallen @zrisher

…--exact] [--graph] [--ceeu] [--dump] Writes to a file the federal income and payroll tax OUTPUT for each filing unit specified in the INPUT file, with the OUTPUT computed from the INPUT for the TAXYEAR using Tax-Calculator. The OUTPUT file is a CSV-formatted file that contains tax information for each INPUT filing unit. positional arguments: INPUT INPUT is name of CSV-formatted file that contains for each filing unit variables used to compute taxes for TAXYEAR. TAXYEAR TAXYEAR is calendar year for which taxes are computed. optional arguments: -h, --help show this help message and exit --reform REFORM REFORM is name of optional JSON reform file. No --reform implies use of current-law policy. --assump ASSUMP ASSUMP is name of optional JSON economic assumptions file. No --assump implies use of static tax analysis assumptions. --exact optional flag that suppresses the smoothing of "stair-step" provisions in the tax law that complicate marginal-tax-rate calculations. --graph optional flag that causes graphs to be written to HTML files for viewing in browser. --ceeu optional flag that causes normative welfare statistics, including certainty-equivalent expected-utility (ceeu) of after-tax income values for different constant-relative- risk-aversion parameter values, to be written to screen. --dump optional flag that causes OUTPUT to contain all INPUT variables (possibly aged to TAXYEAR) and all calculated tax variables, where all the variables are named using their internal Tax-Calculator names. output

codecov-io · 2017-03-20T22:23:16Z

Codecov Report

Merging #1253 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #1253      +/-   ##
==========================================
+ Coverage   99.66%   99.67%   +<.01%     
==========================================
  Files          38       38              
  Lines        2706     2749      +43     
==========================================
+ Hits         2697     2740      +43     
  Misses          9        9

Impacted Files	Coverage Δ
taxcalc/behavior.py	`100% <ø> (ø)`	⬆️
taxcalc/growfactors.py	`98.61% <ø> (ø)`	⬆️
taxcalc/macro_elasticity.py	`100% <ø> (ø)`	⬆️
taxcalc/taxcalcio.py	`100% <100%> (ø)`	⬆️
taxcalc/utils.py	`99.3% <100%> (ø)`	⬆️
taxcalc/__init__.py	`100% <100%> (ø)`	⬆️
taxcalc/calculate.py	`100% <0%> (ø)`	⬆️
taxcalc/records.py	`99.11% <0%> (ø)`	⬆️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 4076f70...6ccbd19. Read the comment docs.

MattHJensen · 2017-03-21T17:59:38Z

taxcalc/cli/tc.py

-                              'assumption file.  No --assump implies use of '
-                              'static analysis assumptions.'),
+                              'assumptions file.  No --assump implies use of '
+                              'standard (std) analysis assumptions.'),


I think "base" might be more appropriate here than "standard" as the word "standard" has normative implications.

Thanks for reviewing #1253. What "normative implications" are you thinking of?

To me, "standard" implies that that a usage is 'approved' and that others are deviant.

For example, consider two Tax-Calculator users on CNN presenting their results. One person could say, "I got X results assuming Y". The second person could say, "Y is just your assumption; when I use the standard model, I got Z."

In this case, Z would sound more reasonable than X.

The same dialogue could take place if the name were "base" instead of "standard", but I think the second guy would have a weaker point.

Something even more neutral than "base" would be preferable, but I can't think of it.

"Static" solved this problem, but I can see how an analysis is still "static" when growdiff_baseline and growdiff_response have been applied.

@MattHJensen said:

Something even more neutral than "base" would be preferable, but I can't think of it.

OK, I see your point. Let me think about alternatives to "standard" and "base".

MattHJensen · 2017-03-21T18:21:36Z

taxcalc/taxcalcio.py

-        growdiff_response Growdiff object used in dynamic analysis;
-        must be None when conducting static analysis.
+        growdiff_response Growdiff object used only in dynamic analysis;
+        must be None when not conducting dynamic analysis.


Not sure whether it is worth splitting hairs here in the code, but growdiff_response must also be None when conducting a partial-equilibrium dynamic analysis.

@MattHJensen said:

Not sure whether it is worth splitting hairs here in the code, but growdiff_response must also be None when conducting a partial-equilibrium dynamic analysis.

I thought your taxonomy was: static, partial-equilibrium(ie, behavior), and dynamic analysis, right? I thought this was the implication of the various quotes you offered last week. If I'm still confused, we need to talk on the phone.

Given that analysis taxonomy, non-dynamic analysis is static or partial-equilibrium, and the wording is sensible.

I thought your taxonomy was: static, partial-equilibrium(ie, behavior), and dynamic analysis, right?

That's not right. My point was that partial-equilibrium analysis should be considered dynamic analysis since it changes GNP, and in the tax-scoring world anything that changes GNP is called dynamic. Will give you a call shortly.

Note that this is the taxonomy we use on TaxBrain since partial equilibrium analysis is available from the dynamic landing page.

From the TaxBrain landing page:

In the context of policy analysis, dynamic modeling incorporates behavior that affects the aggregate output of the economy. This broad definition includes nearly every type of economic analysis except for the approach most often used in the analysis of tax and spending policy proposals by agencies in the federal government.

There is no single "accepted" way to estimate the dynamic effects of policy reforms, so TaxBrain is designed to encourage users to try several approaches and to grapple seriously with the limitations and tradeoffs of each approach.

martinholmer · 2017-03-21T21:45:36Z

@MattHJensen, I'm hoping the three next-to-the-last commits respond effectively to your concerns about terminology in pull request #1253.

martinholmer · 2017-03-22T21:39:12Z

Last call for review of pull request #1253. If there are no concerns raised, I hope to merge #1253 at the end of the day on Thursday, March 23rd.

MattHJensen · 2017-03-23T01:53:02Z

Very excited about the new CLI capabilities. 👍 👍

martinholmer added 11 commits March 19, 2017 15:02

Move tc.py to new taxcalc/cli directory

56aec3e

Check-in tentative meta.yaml changes.

fa8a072

Add taxcalc.cli to packages list in setup.py

ccb5467

Revise meta.yaml to test CLI entry point

24e734e

Simplify tc.py argument checking

6387442

Another change to 'tc --help' output

b553155

Generalize taxcalcio.py logic

fb37c20

Disable pylint import warnings in test_*.py files

2d235f0

Add behavior logic to taxcalcio.py and test it

58e5751

Use tc CLI to conduct file-upload-tests

bc0788b

talumbau added the in progress label Mar 20, 2017

martinholmer added 4 commits March 21, 2017 07:57

Add bash scripts to install/uninstall local taxcalc-cli package

8f6929a

Rename/revise taxcalc/cli package scripts

3b14cb8

Streamline TaxCalcIO class logic

9f42f5b

Add more args value checking to tc.py

5a95d73

MattHJensen reviewed Mar 21, 2017

View reviewed changes

martinholmer added 2 commits March 21, 2017 17:08

Revise dynamic terminology

31c22db

Eliminate 'standar' assumption terminology

3d82a20

martinholmer added 3 commits March 21, 2017 17:49

Further change in replacement for 'standard' terminology

4d9f03d

Change tc OUTPUT filename when no --reform or no --assump

20d9ccd

Remove extraneous pylint comments

6ccbd19

martinholmer merged commit 626387a into PSLmodels:master Mar 23, 2017

talumbau removed the in progress label Mar 23, 2017

martinholmer deleted the taxcalc-cli-dir branch March 23, 2017 22:12

This was referenced Mar 24, 2017

Release 0.8.0 #1226

Closed

Release 0.8.1 #1261

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Make single CLI tool be an entry point to the taxcalc conda package #1253

Make single CLI tool be an entry point to the taxcalc conda package #1253

martinholmer commented Mar 20, 2017 •

edited

codecov-io commented Mar 20, 2017 •

edited

MattHJensen Mar 21, 2017

martinholmer Mar 21, 2017

MattHJensen Mar 21, 2017 •

edited

MattHJensen Mar 21, 2017

martinholmer Mar 21, 2017

MattHJensen Mar 21, 2017

martinholmer Mar 21, 2017

MattHJensen Mar 21, 2017

MattHJensen Mar 21, 2017

martinholmer commented Mar 21, 2017 •

edited

martinholmer commented Mar 22, 2017

MattHJensen commented Mar 23, 2017

Make single CLI tool be an entry point to the taxcalc conda package #1253

Make single CLI tool be an entry point to the taxcalc conda package #1253

Conversation

martinholmer commented Mar 20, 2017 • edited

codecov-io commented Mar 20, 2017 • edited

Codecov Report

Choose a reason for hiding this comment

Choose a reason for hiding this comment

MattHJensen Mar 21, 2017 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

martinholmer commented Mar 21, 2017 • edited

martinholmer commented Mar 22, 2017

MattHJensen commented Mar 23, 2017

martinholmer commented Mar 20, 2017 •

edited

codecov-io commented Mar 20, 2017 •

edited

MattHJensen Mar 21, 2017 •

edited

martinholmer commented Mar 21, 2017 •

edited