Buisness scenario for Predictive Analysis

Hi All,
I am new to SAP Predictive Analysis. I have gone through all the algorithms present in predictive analysis client tool about when should they be used.
Now I planned to try implementing it on some business scenario.
But the problem I am facing is I am not getting any proper data suitable for PA. I am trying to take BW data into excel which will act as my data source.
Can anyone suggest me what data I should pick up from BW or if possible provide me with some sample data.
Thanks and Regards
Mithilesh Prasad

Hi,
You can also found some datasets on SAP Partner Edge website with the Workbook: SAP Advanced Analytics (SAP PartnerEdge).
If you want to find more dataset, I suggest you to download datasets from UCI Machine Learning Repository to build examples. (UCI Machine Learning Repository). You can start with the dataset Statlog (German Credit Data) with a neural network or a decision tree.
Regards,
Edouard Kolf

Similar Messages

How to add a new universe in population for Predictive Analysis

Post Author: izhar
CA Forum: Performance Management and Dashboards
Hi all members I am working on BOXI R2, SP2 I have to ask that, How to add a new universe in population for "Predictive Analysis". Currently i can only view universe of "Direct Customers and Direct Products"
Any member please help me hereRegards Izhar

Hi Anne,
It really depends on how your portal is set up. And there are 101 ways this could have been done.
I've used the Home Page Framework as an easy to define and extend framework to show applications. I've used it both for ESS and MSS, Using the HPF requires that an application has an iView in the portal.
The main point to be made here is that this is not a Web Dynpro ABAP specific question - but rather one about MSS portal implementation - you would probably be better posting a question in either one of the ESS/MSS or Portal forums.
As it is not a WDA specific question, I'd ask that you kindly close this thread and reopen in another forum.
Thanks for your understanding, it would be nice to be able to answer all questions in all forums (perhaps with some sort of tagging system) but currently forum rules are that questions must be specific to the forum posted.
Thanks,
Chris

Need for predictive analysis?

Folks, I have this massive 700,000 sq. ft. facility we need to deploy wireless to using Cisco access points inside the building and also outdoors.
Do you see any need to hire a contractor to do predictive analysis on this?
I mean, I will pay somebody to do the site survey once the building is in place, but I am wondering if typically pay the contractor to execute the predictive analysis instead of just waiting for the real site survey.

A "Predictive Analysis" is not going to guarantee anything. You are better off to wait until the building is build and you have stuff in there befire you hire a company to perform a site survey. Just my thought. An office, might be easier to predict, but not a manufacturing or warehouse facility.

Which R packages are required for Predictive Analysis?

Hi,
I have installed SAP Predictive Analysis and have installed R - 3.0.0.
When I try to use the predict functionality in PA, i keep receiving error messages (such as the one attached).
Is it because I havent installed all of the right packages in R? If so, which packages do I need to install in R?
Thanks in a advance for any help.
Emma

Hi Emma, you can find this doc about configuration of R for previous version with R 2.15 :
How to Manually Install and Configure Open Source R on Microsoft Windows 7 for SAP PA 1.0
Required packages are:
rJava
RODBC
RJDBC
DBI
monmlp
AMORE
XML
pmml
arules
caret
reshape
plyr
foreach
iterator
I have installed these packages with R 3.0.1 with PA 1.19 and it's working.
Hope it helps,
Edouard

Prerequisites for Predictive Analysis ?

Hi Experts,
I desire to be a SAP Predictive Analytics Consultant. I love to play around with data.
So, kindly could anyone tell me how to begin in this path ?
Prerequisites for PA ?
If possible, a small road map for PA ?
Thanks in Advance
Regards,
Gokulkumar RD

Hi Gokulkumar,
Thanks for posting.
I would suggest that you - as you are doing now - look through the articles on this community where you will find good documents that should kick you off in the right direction.
Firstly I would suggest you visit this one:
The magic of Predictive Analytics, HANA & BI.
Secondly here is an end-to-end description of how to use SAP PA & II to solve a predictive task. http://scn.sap.com/docs/DOC-50803
Best regards,
Kurt

Predictive Analysis - Alogorithm determination

Dear Experts,
We are working on a Scenario, and try to build a Predictive Analysis reports on the same, as we are new to HANA , please help us on the same.
Scenario
1) Machine are sending the consumption of liquid to HANA continuously, following are the machines configured.
MachineId
Machine Name
Liquid Flow Limit
UOM
M1
Coffee Wending Machine
2000
Ltrs
M2
Water Purifier
3000
Ltrs
2) Following is the sample data sent from machines for every minute Liquid consumption to HANA.
MachineId
Liquid Dispence
Balance
Date Time
UV
RO
Glight
Red LED
M1
5
1995
4-17-14 12:15 PM
M1
100
1900
4-17-14 12:16 PM
M1
200
1800
4-17-14 12:17 PM
M1
300
1700
4-18-14 12:17 PM
M1
400
1600
4-18-14 12:16 PM
M1
500
1500
4-18-14 12:17 PM
M2
300
2700
4-17-14 12:15 PM
M2
400
2600
4-17-14 12:16 PM
M2
500
2500
4-17-14 12:17 PM
M2
600
2400
4-18-14 12:17 PM
M2
700
2300
4-18-14 12:16 PM
M2
800
2200
4-18-14 12:17 PM
Our Reqirement:
Now we need to Build a Predictive analysis report around this data as follows:
1) The report should show the predictive analysis of liquid consumption by month wise, for each machine level.
Our query:
1) For this requirement what is the predictive analysis algorithm (function) we should use ?
2) What is the implementation procedure on the respective algorithm.
How can we build Types, input Tables ?
How can we build Parameters, how can we determine the Alpha value etc , for our requirement?
Our research and queries on the same.
WE went through a video (HANA Academy - Key Concepts of the Predictive Analysis Library - YouTube)on Single Smooth Algorithm, in that we query on Result.
The input was given as Calendar Id and Sales Amout, where Calendar Id starts from 0 – 23 , what is meant by this Calendar Id ?
And also the result is showing Calendar Id from 2-24 , why it is not considering Calendar 0 and 1
How can we say that the result is predictive and on what basis (like month ,year , date etc).
Thanks and Regards,
Somaraju

Hi JH Yang,
Thanks for the post.
Yes, the BW Connect (offline) feature should be coming to SAP PA.
According to this post: http://scn.sap.com/thread/3571348
Also you find the full Roadmap for Predictive Analysis here: Product Road Map Updates
Best regards,
Kurt

Predictive Analysis 2.0 - 'BW' connect feature ?

Hello folks,
In the latest Predictive Analysis, do we have 'BW' data connect feature ?
BW offline connection or connection via Lumira is fine.
Looking forward to hearing any response. My customer'd like to know this.
Thanks a lot.
JH Yang.

Hi JH Yang,
Thanks for the post.
Yes, the BW Connect (offline) feature should be coming to SAP PA.
According to this post: http://scn.sap.com/thread/3571348
Also you find the full Roadmap for Predictive Analysis here: Product Road Map Updates
Best regards,
Kurt

Can i use SAP Predictive Analysis for this scenario?

Hi guys. I'm new in this subject so I'm a little bit confused about in which cases SAP Predictive Analysis is helpful?
Could it be helpful for this case? I have prices of rice for differente years, from 2007 to 2013. Can SAP Predictive Analysis help me forecast what will be the price of rice for 2014, 2015 and so on?
Which algorithm should I use? Does it come integrated in the tool or is it an add-on?
I´ll appreciate it very much if you could send me the right steps to achieve this.

Hi Erika,
SAP PA has only Triple exponential smoothing algorithm; In most cases, single exponential smoothing would not make any sense [this is available through HANA PAL or R integration with SAP PA];
If you have R integration you should see something like this
In Single exp smoothing only alpha is used;
in double exp smoothing alpha and gamma is used; and all three in gamma.
Why these alpha, beta and gamma... has a big derivation to understand; [let me know if you need any document, I have compiled as on some reference]
What values alpha, beta and Gama should hold in Triple ... or what is the right value, .... actually there are no right value and wrong value for them.
... as for the steps to go in SAP PA, refer below:
Please let me know if you need any assistance ...
Regards,
Arvind E

Predictive Algorithm for Churn analysis

Hi,
Can anybody help me with the algorithm which I can use for churn analysis?
Thanks,
Atul

Hi Atul,
For Churn analysis or what is usually referred to as a binary classification problem where the customer is either staying or leaving=churning I would suggest one of the following algorithms:
CNR Decision Tree - which also provides a decision tree to explain which feature split is influencing the target (churn) the most.
You could also chose one of the R based Neural Network algorithms, however the produced predictive model & results are usually hard to explain.
If need be you can enhance the number of available algorithms by adding you own R functions - there are a lot of examples in this community.
If you have SAP HANA you could also chose:
Decision Trees:C4.5, CHAID or CART (new in SAP HANA SP08).
Other supervised learning algorithms for binary classification: Naive Bayes or SVM (Support Vector Machine).
There are a lot more but this should get you started.
Best regards,
Kurt Holst

Please Provide a Key for SAP Predictive Analysis 1.15 32-bit

please provide a needed a key for SAP Predictive Analysis 1.15 32-bit to be used in University Lab computers.

Hi Ye,
I am afraid but the only way today is to delete your document without PA / Lumira.
By default, local files are stored at :
User/Documents/SAP Lumira Documents
In this folder you'll see the list of all the documents you have created with SAP Lumira or SAP PA.
Delete the document that you don't want anymore and (re)start SAP PA.
The delete functionnality from the desktop tool is an idea accepted by the Labs (Delete Files in Lumira Desktop : View Idea), but not implemented yet.
Hope it helps,
Edouard

Calculationg current and past year for a Analysis report in obiee 11.5

Dear All,
Need some help here.
I am a beginner in OBIEE and need some help to build a report in obiee 11.5 as described below;
I need to create a PnL report such that I show current years profit and loss AND prior year profit and loss figures in a report side by side for comparitive analysis for user.
My data is an essbase database, and have a years dimension with values 'current year' , 'F08', and 'Prior Year'. and other measurs such as revenue, % revenue, gross profit, cost of sales, operating expenses etc. (I am using 107 from oracle downloads) for obiee 11.5
There is another dimesion called time period with members as Q1, Q2, Q3, and Q4, and each having respective months (e.g. Jan, Feb, Mar for Q1 etc)
I also have a scenario dimension with members as Actual, plan, Forecast, Wht -if..
Now I need to create a this report.
I created this report based on approch no 1 below;
Appraoch 1
- Create two analysis reports with all the above columns.
- specify filter for year column equal to current year for first report
- specify filter for year column equal to prior year for second report
- create a dashboard with a promt (radio button) based on column scenario
- create two sections placed side by side below the promt and place above two reports in those two sections so that they appear side by side
Nw when I run this dashboard, it works fine, and shows me the current and prior data for actual, plan etc...
BUT....
I need to create the same reports using a single analysis report instead of two separate reports with hardcoded values for year filter.
Now I tried use AGO funtions on but , it requires a column with numerical value so that it can calculate the past year, but in my case year is a column hving text value and hence i cant use it.
I would like to know how to build such a report using any other variable or function or any rpd level modifitions.
If I had a year dimesion with numeric values, then how would I use to calculate the past year once u select current year .?
Let say values are 2012, 2011, 2010 etc. so If I select 2012 as current year, I should get data for 2012, and 2011, if I select 2011, then I should get data for 2011 ,and 2010.
Please help me to build logic either at analysis level or at repository level
Thanks and Regards
Santosh

Hi,
As per I understand your requirement,
In Pivot table, place 'Year' in pivot columns instead of rows.
Create an SQL filter on Year and put the condition as:
*"Year" BETWEEN @{Presentation_Variable} - 1 AND @{Presentation_Variable}*
You will get the result for selected year and the previous year.
Hope it helps..
Regards,
A.K.

I want a report for Report analysis....

i want a report for Report analysis....
the require ment is like below...
Selection
From Date
To date
Set Up Data
Bank Accounts           FromTo
Billing Accounts         FromTo
% Operations % Management % Research & Development
Above % applies on Billed Amount.
Report Need to be displayed
SBU/Dept     OB Bank     Receipts     Payments     Balance     Accrued OB on Billing     Accrued on Billing     Balance
SBU1
SBU2
SBU3
SBU4
SBU5
SBU6
SBU7
SBU8

Hi Sam
As I said, you can tag variance of FG upto some extent to the COGS of FG...
Also note that, w/o ML, variance is disposed off in the month of mfg, where as COGS comes in the month of sale... For eg: You mfr 100 Kgs @ Actual cost of 120, where as your std cost is 100... You sell 50 kgs in Month 1 and 50 kg in month 2
In Month 1
COGS = 50 (@ std cost)
Variance = 20
In Month 2
COGS = 50
Variance = 0
Variance analysis w/o ML is always limited.. At best what you can do is draw your COPA reports on YTD (Year To Date) basis... Assuming that your closing and opening stock is more or less the same, your variance related to FG can now be analysed with COGS and revenues....
Also, if your prod order for FG is tagged to sale order on account of MTO scenario, your COPA doc for variance will have customer info as well
With ML in place
Variance of SFG can be rolled upto FG... i.e. Say, variance in SFG is 20 USD... Std cost of FG is 125 USD... When you roll up the variance to FG, your FG valuation will become 145 USD....
Also, if part of the FG is sold off, then prorata variance will go to Inventory and the rest can be taken to COGS.. i.e. not only FG, you can also revaluate your COGS with ML...
Regards
Ajay M

Nothing displays on Predictive part in SAP Predictive Analysis

Hi,
Earlier, I created a predictive analysis using R-K Means algorithm, and the result was displayed a week ago and saved it. I opened it several times, trying to beautify the same. But today when I opened it, nothing is seen and also I am unable to add analysis as well. I am using SAP Predictive Analysis version 1.16.0.
Please help me how to resolve the issue.

Hi Anita, in case this has not be resolved, i just had this issue this morning.
I was unable to add or see any Analysis in Predict Tab.
It seems that if you supress your preference settings
(in C:\Users\yourName\.sapvi\predictive_preferences)
including links to all of your documents, and "repairing" Predictive Analysis by Re-installing it, it works again .
It also seem, after reimporting the original issue document, it is blasted, so you'll have to redo it from scratch.
Hope it will work for you, notice me if it does !
Guillaume BUNEL

Oracle Financials or Hyperion Planning for variance analysis??

We are considering Hyperion Planning for our Departmental budgets/forecasts and I'm have a disagreement with a few piers. Was hoping you could help. It's my contention that the budgets/forecasts AND actuals should be populated in Hyperion Planning. A few of my piers contend that the budgets/forecats should be housed in Hyperion Planning and pushed down to Oracle Financials for variance analysis - in order to provide drill back capability into the transacations that caused the variance.
Since H.P. is a consolidated system, I would think it's impossible to budget at a summary account level in H.P. and push it to base level/transaction level within Oracle Financials (one to many). In my view, the actuals should flow up from Oracle Financials, via FDM, into H.P. where variance analysis can take place? I would think drill back capability could be provided via FDM within Planning?
Anyway, was hoping to get your thoughts.

Hi,
We are currently in the final phase of implementing Oracle Financials and Hyperion for the same purpose that you have described in your post. We use Oracle Financials to record actuals and then transfer it to Hyperion under the scenario 'Actual'. While the budget/forecast data in only entered/saved in Hyperion. There are several financial reports for variances that are generated out of Hyperion. So far, this arrangement is serving us pretty well. I second your point that in Hyperion, we can easily compare summary level data. We have Hyperion reports that do variances calculations for sub-totals. In fact, we are also using HP reports for consolidated statements such as Balance sheet, Income statement etc. The actual data for these statements is coming from GL.
As for drill through functionality (from HP to sub-ledger details in financials), we are in process of implementing FDM and use it for both (1) loading data and (2) drill through.
Hope this helps.
ABG

Any pointers to how to do predictive analysis using BI

Hello All
Do you have any leads/pointers about how to do forecasting/predictive-analysis using OBIEE?
I am looking for how to documents or tutorials.
thanks

configure u r ETL and build u r DWH accordingly to the requirements

Buisness scenario for Predictive Analysis

Similar Messages

Maybe you are looking for