WIP: refactor so multiple graphs can be defined for a given slug
WIP: fixed scale view but broke 'All Parent' graph
WIP: working state. All views working properly
WIP: Refactor graph_map into two collections; measure_level_graphs and scale_level_graphs
WIP: refacter GroupedBarColumnPresenter to accept a 'construct' instead of specifying measure or scale
WIP: fix scale graphs being shown on incorrect view
WIP: Merge parents_by_language class with parents_by_language_by_scale so it can handle display of both measure-level and scale-level graphs
Update demographics table with lanugage options
Create a lanugage table to hold the new languages
Update the demographic loader to input languages into the database
Update the cleaner to read the language column
Update the parent table to hold a reference to a language
Update the data uploader script to read the language from the csv and update the language information for any parent items that already exist (or create database entries if none already exist)
update the analyze interface to add controls for selecting ‘parents by group’ and a dropdown for ‘parent by language’
Update the analyze controller to read the parent-by-group parameter
Create a graph for the parent-by-group view
Bubble up averages for language calculations.
Make sure n-size only counts responses for a given measure.
to survey table. We no longer check or keep track of the survey type.
Instead we look in the database to see if a survey item has at least 10
responses. If it does, that survey item was presented to the respondent
and we count it, and all responses when calculating the response rate.
Remove response rate timestamp from caching logic because we no longer
add the response rate to the database. All response rates are calculated
on the fly
Update three_b_two scraper to use teacher only numbers
swap over to using https://profiles.doe.mass.edu/statereport/gradesubjectstaffing.aspx as the source of staffing information
Implement large speed improvements to score calculations. Add page caching to all pages. Small speed improvements to response rate by filtering out
survey items without responses with `none?` `method vs count == 0`.
