Cost History Chart - Weird numbers?

[David Bott]

I have not, but this would mean loosing data.

[Mark42]

I printed a screen capture of the chart, measured the length of each days value bar, and then normalized all the values to the lowest observed. I computed the daily total from the history.db values and then performed the same normalization. The normalized values agreed, within the experimental tolerances.

It appears that the daily values are not stored in the history.db file and are computed in real time when the chart is invoked. The raw values in the history.db file appear reasonable and manually computed daily values are also reasonable.

The chart observations can be explained if (1) the daily values computed are all incorrect by the same percentage or (2) the daily values computed are correct but the single maximum value selected for charting purposes is incorrect.

I have attached two screen cap files. The capture from 13-Jan-11 shows the daily value for 8-Jan-11 as 43.212 (very high, the manually computed value from history.db is a more reasonable 30.688). A week later on 15-Jan-11, the daily value is now shown as 43.212 (also very high, manually computed value from history.db is 33.803) and implying the value on 8-Jan-11 is no longer 43.212.

Mark

[David Bott]

Thanks Mark. Nothing we can do on our end and I would hope that HS would get on this to fix it. After all, they even marketed this product via newsletter and thus I would hope the plugin would work as it should.

[David Bott]

New Chart

Well, just wanted to post this...We surely are not using more Power in January than Dec. The chat is very much a mess.

[Mark42]

David,

I went back and did some more analysis of the history.db data I have and believe I see what is actually happening.

On my chart, both 31-Dec-2010 and 15-Jan-2011 appear to be the same value ~43.212 (a/k/a 100% of scale). Calculating the actual absolute daily totals from history.db with Excel I found that the daily value for 31-Dec-2010 was indeed 43.212 while 15-Jan-2011 was 33.608.

What I believe is happening is that the previous month data (Dec 2010) is being scaled to the Dec-2010 maximum value of 43.212. The current month data (Jan 2011) is being scaled to the current January maximum of 33.608 resulting in both 31-Dec-2010 and 15-Jan-2011 appearing to be 100% of scale (but completely different absolute values).

The reported Y axis scale value of 43.212, which we assume to apply to both the previous month and current month charts, is only valid for one of the charts; in this case the previous month of Dec-2010.

In your data set, I'd guess that you used 83.869 on 1-Dec-2010 and all the Dec 2010 data is scaled to that. The Jan 2011 data is being scaled to whatever the current Jan 2011 maximum (but unknown) value is. On a 0-100% of scale basis, the Dec 2010 is consistent, and the Jan 2011 data is consistent, but the presence of the single Y axis value of 83.869 is misleading.

The charting logic needs to scale both the previous and current month data to the same maximum value, in my case 43.212. Since the single maximum could occur either in the previous or current month, the logic needs to ensure that both sets of data are examined to determine what value to use. An alternative would be to show both months Y axis values, but this would make monthly comparisons very difficult.

Sorry for the wordy reply.

Mark

[David Bott]

well, not that you did the work for them, maybe they can look to fix it. Have you even heard from them on it?

Thanks for the detective work.

[Mark42]

No word but I did update the open ticket with this last bit of (hopefully correct) analysis. I would think the fix would be straightforward.

I assume the analysis made sense with regard to your data set?

[David Bott]

It did.

[Rupp]

Yes this is excellent information but it may still be a bit of time as Rich is working on the memory issues with HSTouch at the moment but thanks a ton for the excellent sleuth work.

[Mark42]

No problem. Glad to help.

[David Bott]

Any updates on fixing this?

Thanks

[Rupp]

The more help desk tickets = more exposure to the problem.

[langenet]

Well... this issue just showed up for me. This morning I looked at my consumption for the day before as usual. The total power consumed for the month was ~22 kwh, yet graphically, it showed 36.092 on the 'This Month vs Last month' graph.

Is there any work/support happening with this plugin anymore?

Robert

[David Bott]

Any updates to this plugin? It is kind of poor not to have real data and even more poor that a user took the time to more or less point out the issue yet no fix.

[langenet]

I placed a helpdesk ticket at the beginning of February - no word yet... I think support for this plugin has taken a back seat.

Robert