search

fslaborg / Deedle

Easy to use .NET library for data and time series manipulation and for scientific programming
http://fslab.org/Deedle/
BSD 2-Clause "Simplified" License
937 stars 195 forks source link

Fix docs for `ReadCsv.inferRows` #453

Closed pkese closed 5 years ago

pkese commented 5 years ago

In the documentation it says that if inferRows is not specified, then all rows will be scanned whereas in the source code, the default is to scan only 100 rows - see the link below for source code.

https://github.com/fslaborg/Deedle/blob/bdf460d851964bef5bbaa37814e8128e12aa9aab/src/Deedle/FrameUtils.fs#L420

I got bitten by this one, because my dataset had only ones and zeros in the first 100 lines so it deduced that the column would be bool instead of int.

In my code, specifically stating inferRows=0 fixed the problem. If the inferRows parameter was omitted, then values other than 0 or 1 were treated as missing.

Similar bug was reported in #271

pkese commented 5 years ago

Travis CI is reporting failing checks, however this pull request is not touching any code whatsoever:
it is only a documentation change.

zyzhu commented 5 years ago

Thanks for the fix. I'll generate new docs in a few days.