Splunk Search

How to edit my search to list top 10 products sold in the last 4 hours, and compare these results to 30 days ago?

demkic
Explorer

Hi folks,

I have Splunk version 6.2.7 and am trying to create a report to display the top 10 products sold within the last 4 hours (in quantity and displayed per product name) and compare those results to the results of 30 days ago.

Ideally, what I would like to be able to see is what are the "new" products (if any) that have been purchased within the last 4 hours that were possibly not purchased 30 days ago. I am trying to find anomalies in products sold.

I am not quite sure my search is displaying any "new" products that may have been sold within the last 4 hours as it compares to 30 days ago.

sourcetype=.... earliest=-30d@d | stats count(product_name) as monthproduct by productname | sort 10 - monthproduct | streamstats count as MonthRank | append [ search sourcetype=... earliest=-4h@h | stats count(product_name) as hourproduct by productname | sort 10 - hourproduct | streamstats count as HourRank ] | stats first(MonthRank) as MonthRank first(HourRank) as HourRank by product_name | eval diff=MonthRank-HourRank | sort HourRank | table product_name, HourRank, MonthRank, diff

Is it appropriate to use append before the subsearch even?

Thanks so much for any insights.

0 Karma
1 Solution

somesoni2
SplunkTrust
SplunkTrust

Give this a try

sourcetype=.... earliest=-30d@d | stats count(product_name) as monthproduct by product_name | sort 10 - monthproduct | streamstats count as MonthRank | append [ search sourcetype=... earliest=-4h@h | stats count(product_name) as hourproduct by product_name | sort 10 - hourproduct | streamstats count as HourRank ] | stats first(MonthRank) as MonthRank first(HourRank) as HourRank values(monthproduct) as monthproduct values(hourproduct) as hourproduct by product_name | eval diff=MonthRank-HourRank | sort HourRank | table product_name, HourRank, MonthRank, diff, monthproduct , hourproduct

View solution in original post

somesoni2
SplunkTrust
SplunkTrust

Give this a try

sourcetype=.... earliest=-30d@d | stats count(product_name) as monthproduct by product_name | sort 10 - monthproduct | streamstats count as MonthRank | append [ search sourcetype=... earliest=-4h@h | stats count(product_name) as hourproduct by product_name | sort 10 - hourproduct | streamstats count as HourRank ] | stats first(MonthRank) as MonthRank first(HourRank) as HourRank values(monthproduct) as monthproduct values(hourproduct) as hourproduct by product_name | eval diff=MonthRank-HourRank | sort HourRank | table product_name, HourRank, MonthRank, diff, monthproduct , hourproduct

demkic
Explorer

That did the trick. My only concern is that the query runs for a really long time now. I am not sure whether this will be manageable in the long-run (for analysis purposes). I did try to break the initial search down as best as I can but it is just really taking a long time. Is there any way to remedy this? Thanks!

0 Karma

somesoni2
SplunkTrust
SplunkTrust

Well, you're querying about 31 days worth of data and it's bound to be slower. See this this work better:-

sourcetype=.... earliest=-30d@d | top 10 product_name showperc=f  | streamstats count as MonthRank | append [ search sourcetype=... earliest=-4h@h | top 10 product_name showperc=f | streamstats count as HourRank ] | stats first(MonthRank) as MonthRank first(HourRank) as HourRank values(monthproduct) as monthproduct values(hourproduct) as hourproduct by product_name | eval diff=MonthRank-HourRank | sort HourRank | table product_name, HourRank, MonthRank, diff, monthproduct , hourproduct

OR

sourcetype=.... earliest=-30d@d latest=now | eval Type=if(_time>relative_time(now(),"-4h@h") ,"Hour", "Month") | top 10 product_name by Type showperc=f | streamstats count as rank by Type | chart values(count) as count values(rank) as rank over product_name by Type | rename "count: Hour" as hourproduct "count: Month" as monthproduct "rank: Hour" as HourRank "rank: Month" as MonthRank
| eval diff=MonthRank-HourRank | sort HourRank | table product_name, HourRank, MonthRank, diff, monthproduct , hourproduct
0 Karma

sundareshr
Legend

There's is a typo in your search, productname vs product_name in the last segement. Other than that, your search should return right results. Why do you believe is missing in the result?

0 Karma

demkic
Explorer

Do you happen to know what can I add extra to the query to display the actual quantities of products sold in addition to what is already displayed? Unfortunately "table product_name, HourRank, MonthRank, diff, monthproduct, hourproduct" doesn't do the trick.

0 Karma

demkic
Explorer

Also, thank you!!

0 Karma

demkic
Explorer

Oh yes, thank you regarding that typo. Actually, it seems as if it is returning the results that I wanted. The only part that it is not able to display on the table for me is the actual quantity of the products for "monthproduct" and "hourproduct" (...when I add in the last command: table product_name, HourRank, MonthRank, diff, monthproduct, hourproduct)

I believe this is because the streamstats command and because I am turning the quantities into rankings. However, how would I be able to tweak the query to also create a column named "monthproduct" and "hourproduct" and display the number of products sold per product?

Many thanks in advance!

0 Karma
Get Updates on the Splunk Community!

Welcome to the Splunk Community!

(view in My Videos) We're so glad you're here! The Splunk Community is place to connect, learn, give back, and ...

Tech Talk | Elevating Digital Service Excellence: The Synergy of Splunk RUM & APM

Elevating Digital Service Excellence: The Synergy of Real User Monitoring and Application Performance ...

Adoption of RUM and APM at Splunk

    Unleash the power of Splunk Observability   Watch Now In this can't miss Tech Talk! The Splunk Growth ...