• check dis box
• uncheck the box above this box
• check dis box too
• definitely check this one and the one below it
• don't check this one
• check this one if you feel like it

nice tool.

wondering if there is a way to view more than two series at the same time? couldnt see in the online demo.

my example dataset

series,timestamp,value,label,
series_a,2019-01-14T16:26:37.000Z,15,
series_b,2019-01-14T16:26:37.000Z,14,
series_c,2019-01-14T16:26:37.000Z,16,
series_d,2019-01-14T16:26:37.000Z,13,
series_e,2019-01-14T16:26:37.000Z,10,
series_f,2019-01-14T16:26:37.000Z,10,
series_a,2019-01-14T16:26:37.100Z,5,
series_b,2019-01-14T16:26:37.100Z,1,
series_c,2019-01-14T16:26:37.100Z,6,
series_d,2019-01-14T16:26:37.100Z,3,
series_e,2019-01-14T16:26:37.100Z,0,
series_f,2019-01-14T16:26:37.100Z,0,

...

This line can cause events that are close together to overlap:

This seems to happen especially when the min_break_sec is small.

Not parsing timestamps with AM/PM properly.

Before, we decided to use primary_threshold as the minimum temperature within a cooking event for it to be considered an event. However, that is not working out well in practice. Let's make a new argument called min_event_temp. Ideally, this argument could be null by default, and the events wouldn't need to have any minimum temperature to qualify as an event.

• add it to S3
• download it when we download the files
• make example code to do this metadata <- read.csv("metadata.csv")
• make sure the project knits
• remove CSV from the repo

When min_event_temp is set to NULL, it is then turned into a numeric value, as.numeric(min_event_temp), so !is.null(min_event_temp) is always true.

• Import SUMs files
• Import metadata
• Merge SUMs and metadata
• Export for TRAINSET
• Create training set
• Import training set
• Calculate features
• Train ML model
• Apply ML model
• Threshold ML results
• Aggregate timeseries data into event-wise data
• Create useful outputs filtered and faceted by metadata

When we import we know:

• sum_make (e.g. Maxim)
• sum_model (e.g. iButon Thermochron DS1922T)
  Let's retain these pieces of metadata

Installation failed on new machine with error:
Skipping 1 packages not available: aws.s3

Manually installing aws.s3 seemed to resolve the issue:
install.packages("aws.s3", repos = c("cloudyr" = "http://cloudyr.github.io/drat"))