Y scale should be based upon observables being displayed, not the entire data set. As example, if lowest temp seen is -20 and highest is 100 and I am looking at graph of temps for the past two days, Y scale should not be -20 to 100, it should reflect the data from the past two days.
X scaling - at a minimum, more data points should be presented than the current max 30 (short term) or 15 (long term). Ideally, like virtually any other app, user should be able to gesture to expand/contract the range with graph adjust size of lines/bars accordingly. Note that the current app does not even present more data if phone is rotated to landscape orientation.
Comment: Disappointed that after latest GUI changes, which included color changes to the graphs, nobody fixed this long standing problem. It really should be trivial to implement.
it isn’t really trivial to implement but indeed a better graphing system is a long standing wish, and something they were supposed to be working on (at least it is on their list). Most definitely there should be more than 30 points on the horizontal axis and it should allow for more flexible zoom levels.
there is a graph for the past day, which scales vertically to the values of the past day. The next zoom level is a graph for five days which zooms to the values in 5 days.
I’ve done android development in the past for financial applications that involved charting model output. It is not very complicated to do the pinch stuff but even skipping that, all charts should be scaled to the same range as the data sent to the charting api. Generally, you would send an array of data points along with a max/min X and Y (if the chart package did not do that on its own).
My shortest multi day graph shows about 3 1/2 days. The range is -20 to 100 while the data points are 60-85. Same happens on the the next level of zoom out (bars). You are corect that on the 15 hour zoom it scales correctly. Again, that should be the case for all levels of zoom.
the data itself often isn’t the problem, but to find an axis with nice looking numbers can be a challenge. I get the impression they implemented their own charting api.
What happens if you scroll to the left in your multi day graph? there is probably you data that shows causes the -20 to 100 scale. It happens to scale for the whole range that you can scroll horizontally for one particular zoom level.
Exactly the issue - the viewable portion of the graph is based on the entire range available rather than the set that is actually viewable. It appears to be based upon only 1 year of history (ironically you can continue to scroll back in time, it just does not plot anything!)
I suppose it is fine if you live in the tropics but any place with more than a 40 or so degree annual range makes for very poor use of the y-axis. That would also apply on shorter term graphs when there are weather extremes.
no, it is not based upon 1 year of history, it is based on the zoom level. If you zoom in to the highest level, you can scroll through one day of data and the vertical scale keeps constant when you scroll. If you zoom out a bit, the vertical axis probably changes as it now scales to 5 days of data (you can scroll through 5 days and the scale keeps the same) . For each zoom level a new scale (unless it happens to be the same ).
Now if that is a good thing, I don’t know, but it does scale.
also that is the most you can zoom out, until it changes to a different style graph. That different graph shows you all the data, with a fresh new scale (but it might be the same in your case, as it depends on the data)
The next level out is the bar style and shows one year. The 3 hour (image above) goes back 6m.
As dew points were below 0F until 3/5, I would not expect the scaling to change unil September and then perhaps to 0/90 (as high has been around 87).
Unfortunately, there were a few days in the upper 40s in early July so the next zoom level is 40/90 throughout.
Unrelated, I’ve just noticed the most recent update appears to no longer get battery data from my Air unit and displays no history, just 0.0V/replace. It is, however, working quite fine and last I recall battery level was fine.
