stocksCRPS and factorsSPGMI sectors misspelled and not official GICS naming conventions used

[spinnj]

factorsSPGMI and stocksCRSP do not use official GICS naming conventions in sector names. In addition, several securities have obvious misspellings in Sector assignment, resulting in incorrect groupings. For example:

unique(factorsSPGMI$Sector) # contains 16 sectors with dupes

[1] "InfoTech" "Industrials" "HealthCare" [4] "ConsumStap" "Energy" "Materials" [7] "ConsumDisc" "TelcoServices" "Utilities" [10] "RealEstate" "Health Care" "Financials" [13] "Consumer Discretionay" "Information Technology" "Consumer Staple" [16] "Communication Services"

For those 16 "Sectors" above, the following mapping would be correct:

Sector (in factorsSPGMI)	Official GICS Sector Name
InfoTech	Information Technology
Industrials	Industrials
HealthCare	Health Care
ConsumStap	Consumer Staples
Energy	Energy
Materials	Materials
ConsumDisc	Consumer Discretionary
TelcoServices	Communication Services
Utilities	Utilities
RealEstate	Real Estate
Health Care	Health Care
Financials	Financials
Consumer Discretionay	Consumer Discretionary
Information Technology	Information Technology
Consumer Staple	Consumer Staples
Communication Services	Communication Services

[spinnj]

One additional note to this is that @braverock has identified the good practice of following official naming conventions for data to facilitate merges with other data sets. I would suggest that in the case of GICS sectors, it it actually better to use the sector number (e.g. "10" instead of "Energy") and then use a function to map numbers to names, since S&P/MSCI have in the past changed the name of a sector (Telecommunications became Communication Services, but the level 1 number remained 50 before and after). Technically this happened in 2018, and the factorsSPGMI data set ends in 2015 so for a true Point-in-Time data set, we should be using Telecommunications instead of Communication Services (the current name for sector 50 since 2018).

[JustinMShea]

I agree @spinnj , the original number would be much better and we can create a function for labels of needed! Along the lines of creating a function to transform the sector number human readable, we could find the dates when name changes happen and store them in a small data.frame that could be incorporated into the function...or simply document them in the /man pages.