Expand my Community achievements bar.

Year over Year alerts

Avatar

Level 2

10/2/18

I would like the ability to set up simple year over year daily alerts. Our industry is very cyclical and/or seasonal on a daily, weekly and monthly basis. Trying to set up day over day, week over week or month over month alerts is not feasible. I have done the analysis on basic metrics; visits, transactions, etc. and I see the variance for the highs and lows for those comparisons as follows;

Day over Day: -47% to +148%

Week over Week:  -51% to +150%

Month Over Month: -71% to +142%

When I do a Year over Year (i.e. -364 days, same day of week) I get the ranges -15% to +21%,

So for us having alerts for a Year over Year day are a much more accurate measure of change and allow for setting tighter thresholds

11 Comments

Avatar

Employee

10/3/18

Thanks for the suggestion - so just to be clear, you're asking for an added granularity of "year" to be added to the alert builder (at the top where it says "Granularity")?

Avatar

Level 2

10/4/18

What I need is to be able to create an alert that compares a metric for yesterday (single day) to the day (single day) 364 days prior. Example

Compare visits for the day Wednesday October 3 2018 to visits for the day Wednesday Oct 4 2017.

I do not need an alert that’s compares an entire year to the entire prior year. Meaning I do not need an alert that compares visits for Oct 4 2017- Oct 3 2018 to Oct 4 2016-Oct 3 2017

Avatar

Employee

10/5/18

Thanks for the clarification - we actually do perform YoY analysis within our Anomaly Detection algorithm, so if you have a daily alert setup to alert on anomalies, it should find those.  If it's not, it would be great to know a little bit more about your data so we can investigate.

Avatar

Level 2

10/5/18

Interesting I wonder why know one made me aware of that given I have been trying to get something in place since last April and ultimately Adobe customer service and engineering both said its not possible to have alerts on a YoY comparison. See ticket

180418-000305 under Sabre Global company

So how would I set up the work space to have the YoY anomaly detection in place and then how to create an alert based on those anomalies?

Avatar

Employee

10/8/18

If you create an alert based on that normal metric, and set it to trigger on anomalies, it should automatically trigger on a YoY anomaly (it checks a bunch of things, including YoY).  So give that a try, and if it is not catching it, please let us know and we can investigate further.

Avatar

Level 2

10/8/18

Ok so I created a very basic alert for Orders in one of our report suites. Ideally I would like to have it alert if there was more than a negative 15% change year over year. I can't set the threshold so I created the alert using a "anomaly is below expected" at a "90% threshold"

According to my data (see below) at a -15% change threshold the alert would have fired once since May 8, 2018. At a -10% change threshold the alert would have fired 4 times since May 8, 2018. And twice in the last 30 days.

I know the anomaly detection is probably a lot more complex than a simple year over year comparison but the Alert work space says the alert would have triggered only 1 time in the last 30 days. Even though the largest negative change (-24.8%) occurred 94 days ago.

So not exactly what I am looking for.

Alert Workspace.png

    

ItemFlowtype Booking (Hit) + PNRs (Bookings) + 7HDXYoY Change
9-May-17224N/A
10-May-17285N/A
11-May-17231N/A
12-May-17224N/A
13-May-17143N/A
14-May-17185N/A
15-May-17263N/A
16-May-17261N/A
17-May-17275N/A
18-May-17246N/A
19-May-17224N/A
20-May-17152N/A
21-May-17209N/A
22-May-17279N/A
23-May-17238N/A
24-May-17232N/A
25-May-17196N/A
26-May-17216N/A
27-May-17149N/A
28-May-17150N/A
29-May-17161N/A
30-May-17255N/A
31-May-17220N/A
1-Jun-17224N/A
2-Jun-17229N/A
3-Jun-17153N/A
4-Jun-17170N/A
5-Jun-17254N/A
6-Jun-17248N/A
7-Jun-17174N/A
8-Jun-17230N/A
9-Jun-17249N/A
10-Jun-17129N/A
11-Jun-17177N/A
12-Jun-17240N/A
13-Jun-17216N/A
14-Jun-17211N/A
15-Jun-17222N/A
16-Jun-17203N/A
17-Jun-17179N/A
18-Jun-17145N/A
19-Jun-17224N/A
20-Jun-17234N/A
21-Jun-17201N/A
22-Jun-17196N/A
23-Jun-17188N/A
24-Jun-17114N/A
25-Jun-17154N/A
26-Jun-17210N/A
27-Jun-17260N/A
28-Jun-17247N/A
29-Jun-17215N/A
30-Jun-17250N/A
1-Jul-17201N/A
2-Jul-17195N/A
3-Jul-17240N/A
4-Jul-17173N/A
5-Jul-17314N/A
6-Jul-17272N/A
7-Jul-17297N/A
8-Jul-17189N/A
9-Jul-17212N/A
10-Jul-17348N/A
11-Jul-17326N/A
12-Jul-17340N/A
13-Jul-17297N/A
14-Jul-17327N/A
15-Jul-17226N/A
16-Jul-17231N/A
17-Jul-17324N/A
18-Jul-17341N/A
19-Jul-17353N/A
20-Jul-17339N/A
21-Jul-17284N/A
22-Jul-17195N/A
23-Jul-17208N/A
24-Jul-17320N/A
25-Jul-17325N/A
26-Jul-17367N/A
27-Jul-17358N/A
28-Jul-17357N/A
29-Jul-17221N/A
30-Jul-17215N/A
31-Jul-17381N/A
1-Aug-17369N/A
2-Aug-17365N/A
3-Aug-17318N/A
4-Aug-17328N/A
5-Aug-17224N/A
6-Aug-17228N/A
7-Aug-17343N/A
8-Aug-17380N/A
9-Aug-17339N/A
10-Aug-17320N/A
11-Aug-17310N/A
12-Aug-17184N/A
13-Aug-17190N/A
14-Aug-17340N/A
15-Aug-17255N/A
16-Aug-17353N/A
17-Aug-17278N/A
18-Aug-17315N/A
19-Aug-17197N/A
20-Aug-17202N/A
21-Aug-17308N/A
22-Aug-17327N/A
23-Aug-17256N/A
24-Aug-17291N/A
25-Aug-17289N/A
26-Aug-17172N/A
27-Aug-17199N/A
28-Aug-17279N/A
29-Aug-17226N/A
30-Aug-17317N/A
31-Aug-17333N/A
1-Sep-17262N/A
2-Sep-17166N/A
3-Sep-17173N/A
4-Sep-17193N/A
5-Sep-17310N/A
6-Sep-17326N/A
7-Sep-17288N/A
8-Sep-17305N/A
9-Sep-17167N/A
10-Sep-17190N/A
11-Sep-17309N/A
12-Sep-17290N/A
13-Sep-17297N/A
14-Sep-17260N/A
15-Sep-17294N/A
16-Sep-17161N/A
17-Sep-17188N/A
18-Sep-17307N/A
19-Sep-17269N/A
20-Sep-17255N/A
21-Sep-17265N/A
22-Sep-17381N/A
23-Sep-17236N/A
24-Sep-17196N/A
25-Sep-17350N/A
26-Sep-17279N/A
27-Sep-17291N/A
28-Sep-17240N/A
29-Sep-17264N/A
30-Sep-17215N/A
1-Oct-17190N/A
2-Oct-17334N/A
3-Oct-17288N/A
4-Oct-17357N/A
5-Oct-17426N/A
6-Oct-17368N/A
7-Oct-17202N/A
8-Oct-17239N/A
9-Oct-17546N/A
10-Oct-17404N/A
11-Oct-17342N/A
12-Oct-17342N/A
13-Oct-17295N/A
14-Oct-17152N/A
15-Oct-17171N/A
16-Oct-17284N/A
17-Oct-17235N/A
18-Oct-17273N/A
19-Oct-17254N/A
20-Oct-17246N/A
21-Oct-17190N/A
22-Oct-17199N/A
23-Oct-17257N/A
24-Oct-17273N/A
25-Oct-17259N/A
26-Oct-17259N/A
27-Oct-17287N/A
28-Oct-17151N/A
29-Oct-17154N/A
30-Oct-17259N/A
31-Oct-17256N/A
1-Nov-17299N/A
2-Nov-17288N/A
3-Nov-17299N/A
4-Nov-17155N/A
5-Nov-17192N/A
6-Nov-17317N/A
7-Nov-17241N/A
8-Nov-17225N/A
9-Nov-17269N/A
10-Nov-17207N/A
11-Nov-17158N/A
12-Nov-17165N/A
13-Nov-17306N/A
14-Nov-17254N/A
15-Nov-17260N/A
16-Nov-17228N/A
17-Nov-17253N/A
18-Nov-17167N/A
19-Nov-17188N/A
20-Nov-17238N/A
21-Nov-17237N/A
22-Nov-17191N/A
23-Nov-1789N/A
24-Nov-17311N/A
25-Nov-17196N/A
26-Nov-17230N/A
27-Nov-17375N/A
28-Nov-17350N/A
29-Nov-17344N/A
30-Nov-17341N/A
1-Dec-17234N/A
2-Dec-17183N/A
3-Dec-17192N/A
4-Dec-17355N/A
5-Dec-17305N/A
6-Dec-17322N/A
7-Dec-17292N/A
8-Dec-17266N/A
9-Dec-17197N/A
10-Dec-17226N/A
11-Dec-17292N/A
12-Dec-17229N/A
13-Dec-17252N/A
14-Dec-17204N/A
15-Dec-17254N/A
16-Dec-17155N/A
17-Dec-17153N/A
18-Dec-17257N/A
19-Dec-17218N/A
20-Dec-17218N/A
21-Dec-17205N/A
22-Dec-17199N/A
23-Dec-17132N/A
24-Dec-1790N/A
25-Dec-17112N/A
26-Dec-17231N/A
27-Dec-17245N/A
28-Dec-17200N/A
29-Dec-17205N/A
30-Dec-17139N/A
31-Dec-17153N/A
1-Jan-18160N/A
2-Jan-18248N/A
3-Jan-18267N/A
4-Jan-18250N/A
5-Jan-18245N/A
6-Jan-18157N/A
7-Jan-18162N/A
8-Jan-18284N/A
9-Jan-18270N/A
10-Jan-18250N/A
11-Jan-18236N/A
12-Jan-18263N/A
13-Jan-18143N/A
14-Jan-18143N/A
15-Jan-18229N/A
16-Jan-18314N/A
17-Jan-18269N/A
18-Jan-18238N/A
19-Jan-18231N/A
20-Jan-18140N/A
21-Jan-18126N/A
22-Jan-18246N/A
23-Jan-18230N/A
24-Jan-18238N/A
25-Jan-18234N/A
26-Jan-18209N/A
27-Jan-18112N/A
28-Jan-18144N/A
29-Jan-18271N/A
30-Jan-18264N/A
31-Jan-18248N/A
1-Feb-18226N/A
2-Feb-18255N/A
3-Feb-18137N/A
4-Feb-18151N/A
5-Feb-18262N/A
6-Feb-18245N/A
7-Feb-18249N/A
8-Feb-18240N/A
9-Feb-18248N/A
10-Feb-18152N/A
11-Feb-18205N/A
12-Feb-18300N/A
13-Feb-18281N/A
14-Feb-18241N/A
15-Feb-18235N/A
16-Feb-18248N/A
17-Feb-18166N/A
18-Feb-18184N/A
19-Feb-18224N/A
20-Feb-18244N/A
21-Feb-18255N/A
22-Feb-18285N/A
23-Feb-18277N/A
24-Feb-18191N/A
25-Feb-18178N/A
26-Feb-18325N/A
27-Feb-18332N/A
28-Feb-18285N/A
1-Mar-18309N/A
2-Mar-18270N/A
3-Mar-18184N/A
4-Mar-18194N/A
5-Mar-18300N/A
6-Mar-18270N/A
7-Mar-18294N/A
8-Mar-18310N/A
9-Mar-18267N/A
10-Mar-18180N/A
11-Mar-18164N/A
12-Mar-18303N/A
13-Mar-18292N/A
14-Mar-18282N/A
15-Mar-18273N/A
16-Mar-18271N/A
17-Mar-18181N/A
18-Mar-18190N/A
19-Mar-18300N/A
20-Mar-18296N/A
21-Mar-18280N/A
22-Mar-18238N/A
23-Mar-18237N/A
24-Mar-18159N/A
25-Mar-18187N/A
26-Mar-18380N/A
27-Mar-18329N/A
28-Mar-18264N/A
29-Mar-18281N/A
30-Mar-18250N/A
31-Mar-18199N/A
1-Apr-18184N/A
2-Apr-18398N/A
3-Apr-18348N/A
4-Apr-18367N/A
5-Apr-18321N/A
6-Apr-18307N/A
7-Apr-18203N/A
8-Apr-18225N/A
9-Apr-18313N/A
10-Apr-18253N/A
11-Apr-18275N/A
12-Apr-18283N/A
13-Apr-18255N/A
14-Apr-18171N/A
15-Apr-18204N/A
16-Apr-18340N/A
17-Apr-18321N/A
18-Apr-18295N/A
19-Apr-18241N/A
20-Apr-18241N/A
21-Apr-18156N/A
22-Apr-18227N/A
23-Apr-18328N/A
24-Apr-18268N/A
25-Apr-18250N/A
26-Apr-18258N/A
27-Apr-18278N/A
28-Apr-18160N/A
29-Apr-18188N/A
30-Apr-18337N/A
1-May-18330N/A
2-May-18280N/A
3-May-18325N/A
4-May-18327N/A
5-May-18143N/A
6-May-18195N/A
7-May-18376N/A
8-May-1834654.5%
9-May-1836126.7%
10-May-1832842.0%
11-May-1829431.3%
12-May-1820845.5%
13-May-181902.7%
14-May-1843766.2%
15-May-1839250.2%
16-May-1834726.2%
17-May-1834540.2%
18-May-1835257.1%
19-May-1823051.3%
20-May-1825421.5%
21-May-1843957.3%
22-May-1835750.0%
23-May-1835954.7%
24-May-1838596.4%
25-May-1840085.2%
26-May-1820940.3%
27-May-1823758.0%
28-May-1823847.8%
29-May-1838249.8%
30-May-1833954.1%
31-May-1830335.3%
1-Jun-1830934.9%
2-Jun-1819326.1%
3-Jun-1821124.1%
4-Jun-1834535.8%
5-Jun-1834940.7%
6-Jun-18355104.0%
7-Jun-1828624.3%
8-Jun-1833032.5%
9-Jun-1819349.6%
10-Jun-1822225.4%
11-Jun-1846794.6%
12-Jun-1839783.8%
13-Jun-1834664.0%
14-Jun-1835660.4%
15-Jun-1838891.1%
16-Jun-1826849.7%
17-Jun-1824266.9%
18-Jun-1837768.3%
19-Jun-1838062.4%
20-Jun-1833767.7%
21-Jun-1832666.3%
22-Jun-1832773.9%
23-Jun-1821286.0%
24-Jun-1822747.4%
25-Jun-1836372.9%
26-Jun-1836941.9%
27-Jun-1839158.3%
28-Jun-1835866.5%
29-Jun-1835040.0%
30-Jun-1822712.9%
1-Jul-1824023.1%
2-Jul-1834945.4%
3-Jul-1830978.6%
4-Jul-18236-24.8%
5-Jul-1832419.1%
6-Jul-1835820.5%
7-Jul-1822016.4%
8-Jul-1825620.8%
9-Jul-1847636.8%
10-Jul-1842028.8%
11-Jul-1841923.2%
12-Jul-1837325.6%
13-Jul-1839721.4%
14-Jul-18223-1.3%
15-Jul-1825510.4%
16-Jul-1849352.2%
17-Jul-1844831.4%
18-Jul-1843222.4%
19-Jul-1844631.6%
20-Jul-1844456.3%
21-Jul-1823621.0%
22-Jul-1827934.1%
23-Jul-1848250.6%
24-Jul-1848348.6%
25-Jul-1846225.9%
26-Jul-1843320.9%
27-Jul-183580.3%
28-Jul-182387.7%
29-Jul-1825920.5%
30-Jul-1848627.6%
31-Jul-1846225.2%
1-Aug-1843619.5%
2-Aug-1841731.1%
3-Aug-1842128.4%
4-Aug-18213-4.9%
5-Aug-1826817.5%
6-Aug-1848541.4%
7-Aug-1849129.2%
8-Aug-1843929.5%
9-Aug-1846144.1%
10-Aug-1837220.0%
11-Aug-1820913.6%
12-Aug-1827544.7%
13-Aug-1852454.1%
14-Aug-1839755.7%
15-Aug-1850543.1%
16-Aug-1840846.8%
17-Aug-1844541.3%
18-Aug-1824222.8%
19-Aug-1829144.1%
20-Aug-1844343.8%
21-Aug-1845037.6%
22-Aug-1846481.3%
23-Aug-1838030.6%
24-Aug-1851377.5%
25-Aug-1821927.3%
26-Aug-1829045.7%
27-Aug-1846365.9%
28-Aug-1835958.8%
29-Aug-183283.5%
30-Aug-183649.3%
31-Aug-1838346.2%
1-Sep-18143-13.9%
2-Sep-18172-0.6%
3-Sep-182066.7%
4-Sep-1844443.2%
5-Sep-1847044.2%
6-Sep-1839236.1%
7-Sep-1841034.4%
8-Sep-1819818.6%
9-Sep-18189-0.5%
10-Sep-1853874.1%
11-Sep-1846460.0%
12-Sep-1837927.6%
13-Sep-1836640.8%
14-Sep-1834718.0%
15-Sep-1820225.5%
16-Sep-1821413.8%
17-Sep-1843541.7%
18-Sep-1841955.8%
19-Sep-1836242.0%
20-Sep-1834931.7%
21-Sep-184138.4%
22-Sep-18205-13.1%
23-Sep-182128.2%
24-Sep-1848137.4%
25-Sep-1842652.7%
26-Sep-1838933.7%
27-Sep-1835949.6%
28-Sep-1840352.7%
29-Sep-18192-10.7%
30-Sep-1823423.2%
1-Oct-1846639.5%
2-Oct-1848468.1%
3-Oct-1846229.4%
4-Oct-1871868.5%
5-Oct-1856353.0%
6-Oct-1826933.2%
7-Oct-1833741.0%

Avatar

Level 2

10/19/18

So a couple of comments, I was merely responding to your comment

"If you create an alert based on that normal metric, and set it to trigger on anomalies, it should automatically trigger on a YoY anomoly"

I set up the work space as stated above and the "Preview" indicated the alert would have triggered 1 time in the prior 30 days. But my data looks as though it should have been triggered twice on a year over year basis.

Secondly, how can I be certain the anomaly flagged is for a year over year comparison as opposed to a possible day over day, week over week or month over month comparison?

It really doesn't seem like it would be that difficult to build in the functionality specifically to have a year over year threshold.

Everywhere I have ever worked (Ecommerce related) we always put more credence and value to a year over year comparison than any other comparison mostly due to the seasonality of our industry over the course of a days, weeks or months. Year over year we see the nearly identical metrics on a daily basis (+/-15% compared to a range of -70% to +150% using any other time comparison)

Avatar

Employee

10/23/18

Thank you for all of this detail.  We do many things in our anomaly detection algorithm, YoY comparison is one of them.  It *should* detect these things for you, and if it's not, we can investigate.  Currently, we don't allow for more granular dissection of the anomaly detection algorithm, but we can certainly consider it (or something like it) for the future.  We will pass this information (your above details) to our data science team to investigate.  Thank you!