Next 4 hour prediction in time series algorithm

I am Sucessed next 7 days prediction,
but now i need to Create next 24 hour Prediction,
how to prepare Data, plz kindly give some suggestion.....
Thanks
Pandiyan
pandiyan

Hi Leo,
Thanks a lot for replay.
but using help of this blog also this problem not solve.
is this problem can we solve using "Seasonal Decomposition of Time Series by Loess".
Regards,
Manish

Similar Messages

SAP HANA One and Predictive Analysis Desktop - Time Series Algorithms

I have been working on a Proof-of-Concept project linking the SAP Predictive Analysis Desktop application to the SAP HANA One environment.
I have modeled that data using SAP HANA Studio -- created Analytic views, Hierarchies, etc. -- following the HANA Academy videos. This has worked very well in order to perform the historical analysis and reporting through the Desktop Application.
However, I cannot get the Predictive Analysis algorithms -- specifically the Time Series algorithms -- to work appropriately using the Desktop tool. It always errors out and points to the IndexTrace for more information, but it is difficult to pinpoint the exact cause of the issue. The HANA Academy only has videos on Time Series Algorithms using SQL statements which will not work for my user community since they will have to constantly tweak the data and algorithm configuration.
In my experience so far with Predictive Analysis desktop and the Predictive Algorithms, there is a drastic difference between working with Local .CSV / Excel files and connecting to a HANA instance. The configuration options for using the Time Series Algorithms are different depending upon the data source, which seems to be causing the issue. For instance, when working with a local file, the Triple Exponential Smoothing configuration allows for the specification of which Date field to use for the calculation. Once the data source is switched to HANA, it no longer allows for the Date field to be specified. Using the exact same data set, the Algorithm using the local file works but the HANA one fails.
From my research thus far, everyone seems to be using PA for local files or running the Predictive Algorithms directly in HANA using SQL. I can not find much of anything useful related to combing PA Desktop to HANA.
Does anyone have any experience utilizing the Time Series Algorithms in PA Desktop with a HANA instance? Is there any documentation of how to structure the data in HANA so that it can be properly utilized in PA desktop?
HANA Info:
HANA One Version: Rev 52.1
HANA Version: 1.00.66.382664
Predictive Analysis Desktop Info:
Version: 1.0.11
Build: 708
Thanks in advance --
Brian

Hi,
If you use CSV or XLS data source you will be using Native Algorithm or R
Algorithm in SAP Predictive Analysis.
When you connect HANA, SAP Predictive Analysis uses PAL Algorithm which runs
on HANA server.
Coming to your question regarding difference,
In SAP PA Native Algorithm, we could provide the Data variable, Algorithm
picks the seasonal information from the Data column. Both R and SAP HANA PAL
does not support Date Column. We need configure seasonal information in
Algorithm properties.
R Properties
1) Period : you need to mention the periodicity of the Data
Monthly : (12)
Quarter : (4)
Custom : you can use it for week or Daily or hourly.
2) Start Year: need to mention Start year.
Start year is not used by algorithm for calculating Time series, but it helps
PA to generate Visualization ( Time series chart) by simulating year and
periodicity information.
3) Starting Period:
if your data is Quarterly and you have data recordings from Q2, mention 2 in
start period.
Example.
If the data periodicity is Monthy and my data starts from Feb 1979, we need to provide following information,
Period: 12
Start year: 1979
start Period: 2
PAL Properties. : Same as properties defined in R.
Thanks
Ashok
[email protected]

Time Series algorithm problem

Hi,
problem-In Time Series Algorithm predication is wrong if Sales value is negative
i have data for more than 3 years ,
sample data like
Year/Month
Zone-1 Sales
201001
-5467.66
201002
2345.88
201003
-7895.33
201004
2345.65
201005
120.2333
Due to negative value my prediction for next month is wrong. Please tell me how I can handle this problem. Negative values come due to sales return in zone.

Hi Leo,
Thanks a lot for replay.
but using help of this blog also this problem not solve.
is this problem can we solve using "Seasonal Decomposition of Time Series by Loess".
Regards,
Manish

How prepare Query for time series algorithm?

Hi every one,
i want next 6 month prediction, how prepare the Query ,
I have Date column,Crime column,Incidents Column,I going with next 6 month so how we get date columns month wise or date wise,
if month wise means,How split the Year and month from date colum??
Please i need some help.....waiting for reply.....
pandiyan

Hi Leo,
Thanks a lot for replay.
but using help of this blog also this problem not solve.
is this problem can we solve using "Seasonal Decomposition of Time Series by Loess".
Regards,
Manish

Add varchar column in to Time Series Algorithm

i created a time series algoritham using numeric columns, but i need to add text columns so what we do?
plz share to me if possible to add.

Hi,
Thanks a lot, it's really help...don't know why I didn't see this customized point before in SPRO.
But I have an other issue, my new fields is added in the CATS screen and now I'm trying to put default value by using the user exit of extension CATS0009.
When I complete the value of my new fields in structure CATSD_IMP, there are not taking into account and the CATS screen doesn't display the value.
Am I using the wrong table ? wrong user-exit ? or my code ..
Thansk a lot again for your help.
Fanny GROUX

How to add new past in mining model and how to update the previous data in the minin model of time series algorithm?

hello,
I creata a mining model and trained it with some data using the sql table.Now I want that when ever I make any Changes in sqltable record. Then I have to chane the data of mining model also.Please provide help.
Thanks in Advance
Shalini Rathore

Hi, thanks for the reply!
Just to follow up for what we did to disable the delete function for maintaining table records. We hided the Delete button by adding a "MODULE disable_delete" code in Screen Painter. So now only adding records to the table is allowed.
Thanks,
Jenny

I need to create Data mining Next 24 Hour Predcition???

Hai Every one...!
am need to Create mining strucuture using time series algorithm ,
i take military hour as a Key time column ,it have time(00 to 23.59), i need the reult is next 24 hours,
i try to create mining structure but it giving result this type
Crime Time Incidents
crime1 24   5
crime2 25
6
crime1 26
7 5
i need next 24 hour means , just assume now time 12 a clock am precocess the mining structure i need to result this type
Crime Time Incidents
crime1 01
8
crime2 02
4
crime1 03
3
plz give any idea,its very usefull for me .....
thanks lot
pandiyan
pandiyan

Hai Every one...!
am need to Create mining strucuture using time series algorithm ,
i take military hour as a Key time column ,it have time(00 to 23.59), i need the reult is next 24 hours,
i try to create mining structure but it giving result this type
Crime Time Incidents
crime1 24   5
crime2 25
6
crime1 26
7 5
i need next 24 hour means , just assume now time 12 a clock am precocess the mining structure i need to result this type
Crime Time Incidents
crime1 01
8
crime2 02
4
crime1 03
3
plz give any idea,its very usefull for me .....
thanks lot
pandiyan
pandiyan

Time Series using different keys

Hi, i'm currently using the time series algorithm to do some forecasting on some data.
My data looks like this :
Date City CustomerID Sales
2003-12-01 New York 12346 45
If i only specify the date as a key time. I can forecast the total sales.(because i am using a cube who aggregates all the data)
But my goal is to forecast sales for each city and/or for each customer.
I tryed to make a cube with date and customer id as keys, but time series isn't compatible with that type of data as a key sequence.
I tryed doing nested tables not working(altough i might do something wrong).
How can use the field date as a key sequence and specify a key like customer id to do forecast the sales / customers?

Hi O G-M,
Each model must contain one numeric or date column that is used as the case series, which defines the time slices that the model will use. The data type for the key time column can be either a datetime data type or a numeric data type. However, the column must
contain continuous values, and the values must be unique for each series. The case series for a time series model cannot be stored in two columns, such as a Year column and a Month column. For more information about it, please see:
http://msdn.microsoft.com/en-us/library/ms174923(v=sql.100).aspx
Thanks,
Eileen
Eileen Zhao
TechNet Community Support

Time Series Question - How to avg for Period children but not parent

I have a calculated account I'm trying to have average at the Period Child level but have the Period Parent level do a straight calc on the totals. My cube is set up like this:
Period
- Jan
-- JanWeek1
-- JanWeek2
-- JanWeek3
-- JanWeek4
-- JanWeek5
Account
- BIlled Hours
- Revenue
- Avg Houly Rate = "Revenue" / "Billed Hours" (I have time series set to average)
My issue is that at the Week level things are averaging correctly, But at the Month level it is taking the average of the 5 weeks. How can I get it to just look at the total Revenue for the month divided by the total Billed Hours (while still having the weeks do an average?)
Thanks in advance for any guidance.

You probably don't need to use Time Balance Average at all for this calculation. I'd suggest setting the member to Time Balance None, then tag it as 'Dynamic Calc', and finally tag it as 'Two-Pass', to ensure that the Revenue / Hours calculation is done after the period roll-up.
I think that would give you the result you want.

Countdown timer to NEXT WHOLE HOUR..?

Hello.
I need to create a countdown timer function.
Normally that would not be a problem for me, but this timer needs to countdown to the next whole hour. As an example, if the time is 23:42 the countdown would say 17:35 (17 minutes, 35 seconds).
When the timer hits zero, it will start from 59:59 and so on.. Always counting to the next whole hour.
Im kind of in a jam and need this done ASAP, any help would be greatly apreciated!
/ Rasmus

Im sorry if i implied that im a great actionscripter, im not
What i would normally do is just grab a counter script from a google search and modify it to countdown to the correct time, but i havent been able to find a counter that counts down to the next whole hour.
It should just be plain and simple and display the time in minutes and seconds like this: 25:14
I found this one on google, counting down for christmas. If that could be modded into showing only minutes and seconds and of course counting down to the next whole hour, that would be awsome!
https://dl.dropbox.com/u/2628604/countdown-test.fla

Best Partition for Time Series

Hi All,
i have the following tables in my DB
CREATE TABLE READING_DWR (
ID     VARCHAR(20)     NOT NULL,
MACHINE_ID     VARCHAR(20),
DATE_ID     NUMBER,
TIME_ID NUMBER,
READING NUMBER
CREATE TABLE DATE_DIMENSION (
DATE_ID     NUMBER     NOT NULL,
DATE_VALUE     DATE     NOT NULL,
DAY     VARCHAR(10),
DAY_OF_WEEK     INTEGER,
DAY_OF_MONTH     INTEGER,
DAY_OF_YEAR     INTEGER,
PREVIOUS_DAY     DATE,
NEXT_DAY     DATE,
WEEK_OF_YEAR     INTEGER,
MONTH     VARCHAR(10),
MONTH_OF_YEAR     INTEGER,
QUARTER_OF_YEAR     INTEGER,
YEAR     INTEGER
CREATE TABLE TIME_DIMENSION (
TIME_ID     NUMBER     NOT NULL,
HOUR     VARCHAR(3),
MINUTE     VARCHAR(3),
SECOND     VARCHAR(3),
INTERVAL     NUMBER
Referential Constrains:-
STG_READING(DATE_ID)>>>>>DATE_DIMENSION(DATE_ID)
STG_READING(TIME_ID)>>>>>TIME_DIMENSION(TIME_ID)
READING_DWR contains the time series data for a particular machine.
What is the best way to partition the READING_DWR to improve the performance of my select query??

Thanks for posting the additional information. I think I have a better understanding of what you are trying to do.
As I suspected partitioning has nothing to do with it.
>
Now where the first value is null , i have to get the record from the READING_DWR , where the time is less then 10:00 for a particular machIne
>
If I understand what you what you are trying to do correctly it is something like this. Please correct anything that is wrong.
1. READING_DWR is a history table - for each machine_id there is a datetime value and an amount which represents a 'total_to_date' value
2. STG_READING is a stage table - this table has new data that will be (but hasn't been) added to the READING_DWR table. All data in this table has a later datetime value than any data in the READING_DWR table. You know what the date cutoff is for each batch; in your example the earliest date is 10:00
3. You need to report on all records from STG_READING (which has 'total_to_date') and determine the 'incremental-value'; that is, the increase of this value from the preceding value.
4. For the first record (earliest datetime value) in the record set for each machine_id the preceding value will be the value of the READING_DWR table for that machine_id for the record that has the latest datetime value.
5. Your problem is how to best meet the requirement of step 4 above: that is, getting and using the proper record from the READING_DWR table.
If the above is correct then basically you need to optimize the 'getting' since you already posted code that uses the LAG (1 record) function to give you the data you need; you are just missing a record.
So where you show output that was from only the STG table
>
Now the output will be
=======================
Time Reading lag
10:00 200 null
10:15 220 200
10:20 225 220
10:30 230 225
>
If you include the DWR record (and no other changes) the output might look like
>
Time Reading lag
08:23 185 null
10:00 200 185
10:15 220 200
10:20 225 220
10:30 230 225
>
The above output is exactly what you want but without the first record. I assume you already know how to eliminate one record from a result set.
So the process for what you need, in pseudo-code, basically boils down to:
WITH ALL_RECORDS_NEEDED AS (
SELECT machine_id, last_record_data FROM READING_DWR
UNION ALL
SELECT * FROM STG_READING
SELECT lag_query_goes_here FROM ALL_RECORDS_NEEDEDThen either ignore or remove the earliest record for each machine_id since it came from READING_DWR and will have a NULL for the lag value. If you add a flag column to each query to indicate where the data came from (e.g. 'R' for READING_DWR and 'S' for STG_READING) then you can just use the records with a flag of 'S' in a report query or outer query.
So now the problem is reduce to two things:
1. Efficiently finding the records needed from the READING_DWR table
2. Combining the one DWR record with the staging records.
For #1 since you want the latest date for each machine_id then an index COULD help. You said you have an index
>
index on READING_DWR---MACHINE_ID,DATE_ID,TIME_ID
>
But for a query to find the latest date you want DATE_ID and TIME_ID to be in descending order.
The problem here is that you have seriously garbaged up your data by using numbers for dates and times - requiring
>
TO_DATE(DATE_ID||''||LPAD(time_id,6,0),'YYYYMMDDHH24MISS'))
>
to make it useful.
This is a VERY BAD IDEA. If at all possible you should correct it. The best way to do that is to use a DATE column in both tables and convert the data to the proper date values when you insert it.
If that is not possible then you should create a VIRTUAL column using your TO_DATE functionality so that you can index and query the virtual column as if it were a date.
For #2 (Combining the one DWR record with the staging records) you can either just union the two queries together (as in my psuedo-code) or extract a copy of the DWR and insert it into the staging table.
In short query ALL of the DWR records you need (one for each machine_id) separately as a batch and then combine them with the STG records. Don't look them up one at a time like your posted code is trying to do.
If your process is something like this and perhaps run every 15 minutes
1. truncate the stage table
2. run my report
3. add stage records to the history table
Then I would modify the process to use the 15 minutes 'dead' time between batches to extract the DWR records needed for the next batch into a holding table. Once you do step 3 above (update the history table) you can run this query and have the records preprocessed for your next batch and report.
I would use a new holding table for this purpose rather than have the staging table server a double purpose. You never know when you might need to redo the staging table load; this means truncating the table which would wipe out the DWR staged records.
Anyway - with all of the above you should be able to get it working and performing.

TODATE time series function in OBIEE 11g

Hi,
I have a problem with time series function. I have month level in time dimension hierarchy.
I have used below expression to get month to date results in my reports.
column
expression----> TODATE(fact.measure, hierarchy.Month level);
when i am using this column in my reports it is showing null values. The below error i am getting in view log files
----------------> Converted to null because it's grain is below query's grain
Note: Here i have measures, year,qtr,month,day,shift,hour in single physical table in physical layer.
Is it a problem to have measures and time columns in a single physical table?
Please let me know if you have any solution.
Thanks,
Avinash

Yes, it shud be a prob. Try using seperate tables for fact n timedim

Regarding Time Series Graph in OBIEE 11g

Hi,
I need to create a time-series graph in OBIEE 11g. However, the value for time on the x-axis of the graph automatically comes for days.
Is there any way to change it to hours or minutes?
Thanks,
Naman Misra

Yes, it shud be a prob. Try using seperate tables for fact n timedim

Time Series Plot

Is there a standard way to create a time series chart in Apex 4.2?
The line chart does not take into account irregular time intervals - it assumes that the data points on the X axis are equally spaced, its just really a bar chart but with a line instead.
The scatter chart handles the x axis values correctly but does not have the option to join the points with a line, so all sense of time series trend is lost.
Using the line chart I can fudge the values to add extra null values for a defined list of dates so that the spacing of points is more accurate, but this is only really ok when say there is one point per day and there may be missing days. If there are sometimes 3 points per day or zero points per day this approach is too cumbersome.
Alternatively I can change the scatter chart to use a line. Anychart has the option to use a line (line_style) but this isn't included in apex and it cant be added to the XML because it exists within the #DATA# section so the only way would be to fudge the series query to bring that back before the data.
I would have thought that a time series plot would be quite a common requirement - have I missed it somewhere?
Kathryn

Hi thanks for the reply.
I have attached the report I have made. I have written a VBA code to fill in all the details and graphs accordingly. Now the test requirements are as given in the table. I need to specify that in the graph. I dont want the encircled straight line. Just the 2 limits from 10 to 30 ms. In fact the more correct and logical wasy of representing the limits would be 3 horizontal lines, 1 at 10 ms from 22.5-27.5g, 1 at 20 ms from 17.6-22.6g and 1 at 30 ms from 12.5-18.5g.
Hope i have put my doubt across clearly. Again with the code you have given, I dont know how to 'tell' diadem to plot that particular graph in that particular X-Y plot in the report/LPD.
For the 2 lines, this is the code i have written.
'Find 10 ms time
        L3= ChnLength(1)
        L4 = 0
        Do while (L4 < L3)
        L4= L4 + 1
        R1= ChD(L4,1)
          If (R1 > 0.01) Then
                Exit Do
          end if
                 Loop
        L4= L4 - 1
        LT10=L4
'Find 20 ms time
        Do while (L4 < L3)
        L4= L4 + 1
        R1= ChD(L4,1)
          If (R1 > 0.02) Then
                Exit Do
          end if
                 Loop
        L4= L4 - 1
LT20=L4
LT1 = LT20-LT10
          For i = 0 to LT1 'Lt1 is var for channel length from 10 ms to 20 ms
            ChD(LT10+i,"Upper limit") = 27.5-(4.9*i/LT1)
          Next
Thanks again. Waiting for your kind help!
Attachments:
neck cal.jpg ‏150 KB

Time series chart doesn't work

I am trying to create a simple time series chart, plotting five time series over a 5 year period. I have a table with the time series in rows, but when I select the chart to graph it, the data appears as if the time series is in the columns! Then, when I re-created the chart with the data in the columns and selected it, it created the same chart! I tried "transposing" the columns and rows as the "Help" command suggested, by going into the "Data Series" box, but it won't accept any input in the box. Help.

Hello
(1) Before charting time or duration values, I convert them as decimal numbers representing a decimal number of days (you may choose a decimal number of hours or a decimal number of minutes).
Here is what I get.
It appears that the chart tool defaults to the orientation where there is the greater number of values.
This is the case for Graphique 1 and for Graphique 2
Clicking on the orientation button I may change the behavior accordingly as I did for Graphique 1 rotate and Graphique 2 rotate.
I don't see any oddity here.
Just for info, to build the transposed tables, I didn't used the TRANSPOSE() function but formulas like:
B9 =OFFSET($D$2,COLUMN()-COLUMN($B$9),ROW()-ROW($B$9))
fill down, fill right
G9 =OFFSET($I$2,COLUMN()-COLUMN($G$9),ROW()-ROW($G$9))
fill down, fill right
Yvan KOENIG (from FRANCE mercredi 28 janvier 2009 14:29:02)

Next 4 hour prediction in time series algorithm

Similar Messages

Maybe you are looking for