Note:

• PROC REG ist used to calculate the linear trend over time. This is a very efficient method even on large datasets (millions of observations)
• Do not forget to specify the NOPRINT option in PROC REG
• You can specify time horizons for the calcuation with a WHERE-statement
• For segmentation and prediction purposes it is very usefull to create a Group Indicatior variable

This sample was authored by Gerhard Svolba. His book, Data Preparation for Analytics is available from the SAS Publishing online bookstore.

---

These sample files and code examples are provided by SAS Institute Inc. "as is" without warranty of any kind, either express or implied, including but not limited to the implied warranties of merchantability and fitness for a particular purpose. Recipients acknowledge and agree that SAS Institute shall not be liable for any damages whatsoever arising out of their use of this material. In addition, SAS Institute will provide no support for the materials contained herein.

*** Step 1 - Create Sample Data;

data wide;
 input CustID M1 M2 M3 M4 M5 M6 8.;
 Cards;
1	52	54	58	47	38	22
2	22	24	30	28	31	30
3	100	120	110	115	100	95
4	43	43	43	.	42	41
5	20	29	35	39	28	44
6	16  24	18	25	30	24
7	80	70	60	50	60	70
8	90	95	80	100	100	90
9	47	47	47	47	47	47
10	50	52	0	50	0	52
 ;
run;

PROC TRANSPOSE DATA = WIDE OUT = LONG;
 BY custId;
RUN;

DATA LONG;
 SET LONG;
 FORMAT Month 8.;
 RENAME col1 = Usage;
 Month = compress(_name_,'M');
 DROP _name_;
RUN;

*** Step 2 - Create Measures for Course over time with Proc Reg;

*** a) Long-Term Trend - all six months;

PROC REG DATA = long NOPRINT 
         OUTEST=Est_LongTerm(KEEP = CustID month
                             RENAME = (month=LongTerm));
 MODEL usage = month;
 BY CustID;
RUN;


*** b) Short-Term Trend - only last3 months;

PROC REG DATA = long NOPRINT 
         OUTEST=Est_ShortTerm(KEEP = CustID month
                              RENAME = (month=ShortTerm));
 MODEL usage = month;
 BY CustID;
 WHERE month in (4 5 6);
RUN;

DATA mart;
 MERGE wide
       est_longTerm
       est_shortTerm;
 BY CustID;
 Format ShortTerm LongTerm 8.1;
RUN;




PROC FORMAT;
 VALUE est LOW -< -1     = '-'
           -1  -  1      = '='
            1   <- HIGH  = '+';
RUN;

DATA mart;
 SET mart;
 Format LongShortInd $2.;
 LongShortInd = CAT(put(LongTerm,est.),put(ShortTerm,est.));
RUN;

---

These sample files and code examples are provided by SAS Institute Inc. "as is" without warranty of any kind, either express or implied, including but not limited to the implied warranties of merchantability and fitness for a particular purpose. Recipients acknowledge and agree that SAS Institute shall not be liable for any damages whatsoever arising out of their use of this material. In addition, SAS Institute will provide no support for the materials contained herein.

Cust                                                            Short      Long
 ID      M1     M2     M3     M4     M5    M6    LongTerm        Term    ShortInd

  1      52     54     58     47     38    22        -6.0       -12.5       --
  2      22     24     30     28     31    30         1.7         1.0       +=
  3     100    120    110    115    100    95        -2.3       -10.0       --
  4      43     43     43      .     42    41        -0.4        -1.0       ==
  5      20     29     35     39     28    44         3.5         2.5       ++
  6      16     24     18     25     30    24         1.9        -0.5       +=
  7      80     70     60     50     60    70        -2.6        10.0       -+
  8      90     95     80    100    100    90         1.0        -5.0       =-
  9      47     47     47     47     47    47         0.0         0.0       ==
 10      50     52      0     50      0    52        -2.7         1.0       -=

Operating System and Release Information