Future sales Forecasting

KNIME Analytics Platform
1

Hi, I am new in KNIME and I want to develop a forecasting data mining model to predict seasonality and trend. I have data set like

 

Date                    ITEM_KEY         ORDER_QTY    LINE_AMOUNT (ORDER_QTY*Sale Price)

21/03/2013           I1                      20                            60

22/03/2013           I1                       30                           90

23/03/2013          I1                       5                               15

24/03/2013          I1                       2                                6

 

…….

20/03/2014         I1                        25                              75

21/03/2014         I1                         21                             63

22/03/2014        I1                          3                                9

23/03/2014        I1                          4                               12

 

In my data weekend is always low sale rather than working day because weekend customers order very less. As for example 23/03/2013 -24/03/2013 is the weekend sale and same 22/03/2014- 23/03/2014 also weekend sale and If we compare 22/03/2013 (business day) sale is 30 items but same date 22/03/2014 (weekend) sale is 3.

Similar Christmas sale, mother’s day sale, valentine sale data exist in my data set.

From my data set (last 3 years data) I want to predict the future sale for 24/03/2014 - 10/04/2014 daily basis sales forecast. i.e.

Date                ITEM_KEY                      Order_QTY

24/03/2014              I1                               ?

25/03/2014              I1                            ?

26/03/2014               I1                           ?

27/03/2014                I1                            ?

28/03/2014                 I1                          ?

 

Please can anyone provide me the guidance how to do it in KNIME and it will be great if someone provide me a workflow as well.

Thanks.

2

Hello, 

This is currently easiest to do using our R integration.  We are planning on introducing some native AIRMA support in KNIME in the near future, but I can't comment on any details about that yet.  

For the time being, I find the guide linked below very helpful for understanding how to do the type of analysis you are mentioning in R. Hopefully this helps.   Also, the lag node in KNIME will get you part of the way there in KNIME already.  

http://a-little-book-of-r-for-time-series.readthedocs.org/en/latest/src/timeseries.html

Regards,

Aaron

3

Hi Aaron:

Any news about introducing some native AIRMA support in KNIME?

Thanks.

Gabriel.

4

Hi Aaron,
Any news about the ARIMA nodes?