Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Tree: 965495451a
Fetching contributors…

Cannot retrieve contributors at this time

38 lines (33 sloc) 1.903 kB
* INTC summary has crazy growth rates. Need to re-calc in FSS?
* NVDA summary makes it look like a victim of base multiplier bug
* traversing upward for base multiplier is sometimes going out of control and
finding matches way below the table (e.g., ALOT should be 1 but is getting 1000)
* Assets training examples include garbage like "(in millions)", "assets",
"current assets:", etc.
* Not able to handle reports that are in the exhibits
* Not able to handle reports that have tables split across pages
* Cache aging (refresh at some point)
* Update the classifiers to do word digrams/trigrams, to see if word ordering
improves upon accuracy. Use other features, too, like # of rows in the
current section, so far; absence/presence of values; etc
* Parse "Net income attributable to noncontrolling interests" (which occurs
after the "Net Income" line). See, e.g., Yahoo's 2011 Q1 (Mar 31) 10-Q
Test Quality/Coverage
* Increase the number of companies/years tested in balance sheet specs
Code Quality
* FSS should be in dollars, not $000s
* Think through whether merge should use original year's reported values, or
their possible restatement in later years
* Shift some calculations from QuarterlyReport to FinancialStatementSummary, so
that they'll be accurate post-merge
* Refactor a common parent class for the asset/liab/equity classifiers
* Get rid of remaining absolute pathnames
* Add an install script (download test vectors, etc, etc)
Jump to Line
Something went wrong with that request. Please try again.