fiendskrah/finalcapstone

reinitialized 5/14

finalcapstone's Issues

100 is missing data

[Errno 2] No such file or directory: 'clean_data/lcff.csv'

exported environment.yml will not work on Linux machines.

Preliminary googling seems to indicate that the error is a result of exporting an environment.yml file from Windows to Linux. Upon running conda env create -f environment.yml, I get a short failure message with little guidance, but googling keyphrases seems to indicate that the file cannot be used to create the environment because certain package dependencies are 'windows only'.

Esda bug with local Moran's i

When attempting to calculate local Moran's, esda returns the following error:

"ValueError: cannot assign slice from input of different size"

See 400

Cell 06: constructing queen weight matrix
Cell 14: setting variable
Cell 23: attempting to calculate local Moran's i
@sjsrey

Index project

Data work and analysis for MPP entirely contained within project.ipynb and Analysis.ipynb. We can split the work up into smaller, more digestible notebooks. This makes data work easier to reference later and will allow the project to be easier to adapt to other purposes. Use numerical naming convention.

selecting districts for regression discontinuity design

Beginning investigation in RDD.ipynb. Considering that we still need both treatment and control districts:

1. What criteria should be used to determine treatment vs. control districts?

• Breakpoint approach makes the most sense, as the policy being evaluated has a strict implementation point (upc > 55%)

2. Does it make sense to also evaluate the most disadvantaged districts (upc >~80%) vs. the most advantaged districts (upc < ~20%)? Is this a valid approach?

3. How can we use geosnap/segregation for this project?

• geographically select districts?
• dissolve income data from geosnap census tracts into school districts?

understand missing grade data

in 500_matching.ipnyb:

When slicing 'master' dataframe for df[df['e_curr ALL'].isnull()], pandas returns 120 records with no grade data. I would like a list of these districts to then return to and query against the original unmerged ela and math datasets to ensure no grade data is being erroneously discarded.

check robustness of t-test/p-value function

in cell 9 of 310.

The function runs, but all tests in the following cells return p-values of 0.0.

• Test the function on a new set of play data to ensure its validity.

Create a better README for 'front' of project

README should:

• give a concise description of the policy (CAL SB-97 Concentration grant)
• discuss research questions
• explain applicability to other projects (use this as a reference guide for other policy analyses)
• explain workflow
  • conda environment
  • notebook flow

pandas multi-indexing

One issue here is that I've had trouble merging both english and math scores onto one row of data corresponding to a specific student group within a district. I've tried left, right, inner, and outer merges.

I'd like the dataframe to be structured like so:

• County
  • District
    • Student Group
      • English Score
      • Math Score

Instead, the student group gets duplicated for each set of scores.

• County
  • District
    • Student Group
      • English score
    • Student Group
      • Math score

I got around this for the capstone by simply duplicating the analysis for a second dataframe.

• County
  • District
    • Student Group
      • English score

and

• County
  • District
    • Student Group
      • Math score

Compare workflow to NARSC'20 workshop

found here and on your /d/ drive.

See particularly notebooks 200 - 500