Protea_Kudu forecasts have unexpected indentation in the ‘value’ column

FluSightNetwork / cdc-flusight-ensemble

Guidelines and forecasts for a collaborative U.S. influenza forecasting project.

http://flusightnetwork.io/

37 stars 22 forks source link

Protea_Kudu forecasts have unexpected indentation in the ‘value’ column #233

Closed Khoale1096 closed 4 years ago

Khoale1096 commented 4 years ago

I noticed some forecasts of Kudu have tab indentation in the 'value' column like in the image:

According to the common format of csv https://tools.ietf.org/html/rfc4180#page-2, "Spaces are considered part of a field and should not be ignored.". Therefore when the scripts read this value field, they get values with leading white spaces, rather than just the float numbers. Therefore I think this should be fixed. I've seen this issues in the real-time forecasts this season, as well as the retroactive forecasts of last season for Kudu.

craigjmcgowan commented 4 years ago

I believe this is fixed for the EW44 forecasts - it's a result of coercing the Value field to character in R as a result of #215.

I just used trimws in R on the Value field. Do you need me to adjust the training forecasts as well?

nickreich commented 4 years ago

I wondered if it wasn't tied to the other formatting issue. If it's not too much trouble, would be great if you could adjust the training forecasts as well, and the first few real-time ones for this year.

craigjmcgowan commented 4 years ago

I can do that. May be a few days before I have time but will update before next week's submission