My Acurite Charts: using separate continuous lines for temperature highs and low just doesn't make sense

  • 1
  • Idea
  • Updated 2 years ago
Archived and Closed

This conversation is no longer open for comments or replies and is no longer visible to community members. The community moderator provided the following reason for archiving: Archiving due to lack of activity

In My Acurite's chart view for anything over three days shows two separate continuous data lines for temperature high and temperature low.   But the data isn't continuous.   It's just data points that cover some intervals/buckets.  But there are these misleading lines.  Worse, the lines between the actual data points appear to have been made into curves to make them look better.

The visualization is just wrong, the low temperature didn't vary over a smoothed curve during that time.   It looks to me like the only actual data in the chart are the daily data points.  So why are they connected by lines?  

The data should be represented by a more appropriate and more accurate graph type,  Scatter plots, or stacked bar charts, would make a lot more sense.
Photo of Rob Terzi

Rob Terzi

  • 73 Posts
  • 8 Reply Likes

Posted 2 years ago

  • 1
Photo of Bill Martin

Bill Martin

  • 758 Posts
  • 127 Reply Likes
Myself and others have posted similar thoughts about these graphs.  What is really bad is when gaps (time stamps that had LoS/LoC) are shown to have valid data due to the graphing technique.  The only way you can really see the actual data is to download a CVS file and graph on your own.  

The current graphs are very misleading.  
Photo of Rob Terzi

Rob Terzi

  • 73 Posts
  • 8 Reply Likes
Thanks Bill.   I figured others had complained but I didn't find the threads to add my support to via Me To / Follow.

I've posted a bunch about misleading and just plain erroneous data during loss of signal and/or loss of of connectivity.  The iOS app shows some clear problems where it looks like uninitialized data is used, so the data you seen when there is data missing depends upon what screen you were on before.
Photo of Bill Martin

Bill Martin

  • 758 Posts
  • 127 Reply Likes
Rob: thanks for doing different types of testing. Extremely bad. Doing practice along with unit/integration/alpha testing processes.

Uninitialized or taken previous (very old data) is not good.

Also depending. How you access a graph (ie temperature first and then going to pressure, the graph will truncate the pressure so the axis labels are not fine enough). Each time a specific graph is chosen, the correct format should be set rather than assuming the previous format is sufficient for new data. This is apparently on phone applications. On the PC MyAcurite.com, all the graphs are shown at one time and show correct axis labels. Phone, you need to select one graph at a time from scroll down menu.
Photo of Sam Kale Jr.

Sam Kale Jr.

  • 9 Posts
  • 2 Reply Likes
Maybe it would be better to provide the option to have an option of line, scatter, or bar chart...
Photo of Bill Martin

Bill Martin

  • 758 Posts
  • 127 Reply Likes
Sam: yes, Chaney needs to look at their graphing tab and how it can be much more user friendly. It all depends on their development sw package and what it allows to be implemented.

- How graphs are done can be user selectable

- Colors used (10-15% of population is color blind and it is not just red/green.

- Turn on/off multiple data exists but is not persistent. Let's say you get to a graph, deselect data you don't want and then decide to view a different time period. All is reset. This should selectable and once set, does not change until user changes it. This is not just during a current session but if you log in a few days later. After all, Chaney knows it's you since it is your login and data being viewed. Redoing this everytime is ridiculous and poor design from a user's perspective.

I would encourage Chaney Engg to read "The Inmates are running the Asylum". All about user interfaces. It talks about Use Cases (end users) and making interfaces intuitive. Chaney has done some of this by the tabs (Dashboard, Graphs, setup, etc) and grouping like functions into areas. But in each tab, I believe they could have done more.

There are also some other excellent form vs. function books (ceramic pottery) that discuss from different, non software based perspectives.
Photo of Sam Kale Jr.

Sam Kale Jr.

  • 9 Posts
  • 2 Reply Likes
So another option would be to download cvs file, build chart/graph or dashboard presentation which could be automated, then have an option in My AcuRite to upload a snapshot of your chart into a new tab
Photo of Rob Terzi

Rob Terzi

  • 73 Posts
  • 8 Reply Likes
Sam thanks for your suggestion.   However, the current CSV interface would be difficult to automate for many people.  The process is to request the CSV export, then wait for the email that contains the link.

Instead it would be easier to just get the data using another method such as captuing the aculink packets or using a separate RF receiver.  There are several working solutions that have been published.
Photo of Bill Martin

Bill Martin

  • 758 Posts
  • 127 Reply Likes
Sam: lots of options that Chaney can choose from. Some probably more viable based on Chaney's development environment.

I hope Chaney views this from an end user's perspective which includes accuracy as well as ease of use.
Photo of AcuRite Jennifer

AcuRite Jennifer

  • 11801 Posts
  • 536 Reply Likes
Hello All,

Thank you for the feedback we will look into this for future product development.



Rob- When you chart up to 24 hours the chart line will show in 5 minute intervals, up to 3 days will show 15 minute intervals, and 4-31 days will show 1 reading per day with the high and lows.  If you are charting 3 days Rob you should get 15 minute intervals.This would be charting from November 1st - November 3rd for example. You stated you charted 2 day and it shows the high and low. It should provide you with more information then that. Please try the recommending charting and let me know if you see more than the high and low.  Thank you. Have a good day!
(Edited)
Photo of Rob Terzi

Rob Terzi

  • 73 Posts
  • 8 Reply Likes
@Acurite Jennifer, thanks for the response.  Please note my first sentence in this thread says, "In My Acurite's chart view for anything over three days shows two separate continuous data lines for temperature high and temperature low.   

I may have stated things confusingly, but I was referring to there being only 1 set of data points per day when the range is 4 days or more.     If there is only 1 set of data points per day, used to generate the graph, then why are the lines between the data points curved?   Is that based on actual data, or was a curve fitting algorithm used to make it look nice?    If the later, the problem is it gives the misleading impression there is real data there, when there isn't. 
Photo of AcuRite Jennifer

AcuRite Jennifer

  • 11801 Posts
  • 536 Reply Likes
Hello Rob,

The lines are there more to make it look nice there is no significance to them. Thank you. Have a good day!
Photo of Bill Martin

Bill Martin

  • 758 Posts
  • 127 Reply Likes
It is curve fitting where the actual min and max are not the min/max in the graph. The actual are within the curve but not shown as min/max
Photo of AcuRite Rachell

AcuRite Rachell, Employee

  • 6411 Posts
  • 200 Reply Likes
Hello All,

We appreciate the feedback and will certainly keep this in mind for possible future updates. Thank you and have a great day!
Photo of Bill Martin

Bill Martin

  • 758 Posts
  • 127 Reply Likes
Rachell:

I was pleasantly surprised when I accessed Charts from my phone and noticed that each chart had the correct number of significant digits for that metric. In previous sw versions, you could get pressure to have the same y axis labels.
So one of the cloud updates must have addressed this. Thanks!

I'll have to examine these charts for min/max and see if the curve fitting algorithm was also changed so the graph's min/max coincide with the actual measured min/max.

Bill
Photo of Bill Martin

Bill Martin

  • 758 Posts
  • 127 Reply Likes
Just checked min/max and the chart's min/max are still different than the actual measured min/max. Chaney is using some algorithm to perform the curve smoothing, so Chaney is using all of the data to curve fit and will regardless of which algorithm is chosen. Issue is the current algorithm is not accurate.

But at least one aspect of charting is improved.
(Edited)
Photo of Rob Terzi

Rob Terzi

  • 73 Posts
  • 8 Reply Likes
I can't see how using a curve to show 7 discrete data points is going to show anything other than garbage information.
Photo of Bill Martin

Bill Martin

  • 758 Posts
  • 127 Reply Likes
Choose a month, same issue. We had a product that allowed users to select between actual or smoothed graphs. Smoothed had actual min/max.

This conversation is no longer open for comments or replies.