Estimating weights each 20 days using data from other ages in dogs

[jonatan_velarde]

Good night my friends:

Nice to see you again, here i go with an exercise that i hope it to be possible to solve. Let get it on!

There has been studied 4 dogs in relation to their growth. This study was related to their weights and its evolution during time. So, I have data from their weights as follows in the table 1:

TABLE 1. RAW OBSERVED DATA

DOG	AGE_IN_DAYS	WEIGHT
1	0	3.5
1	19	4.8
1	47	11.3
1	75	22
1	103	21.5
1	131	34.8
1	159	32.3
1	190	37.1
1	215	42.4
1	243	.
1	272	.
1	299	.
1	327	.
2	0	3.7
2	23	7.7
2	51	14.3
2	79	18
2	107	23.2
2	135	25
2	163	29.3
2	194	33.9
2	219	37.6
2	247	.
2	276	.
2	303	43.2
2	331	43.0
2	359	43.5
2	387	.
2	415	.
3	0	4.9
3	23	6.1
3	50	13.2
3	78	17.2
3	106	20
3	134	25
3	162	28
3	184	29
3	217	32.8
3	246	34.8
3	274	37.4
3	302	36.4
3	330	38.6
3	358	40.4
3	386	37.8
3	414	37.4
3	442	37.2
3	470	35.4
3	498	36.1
3	526	42
3	554	40.4
3	786	64.4
3	814	65.9
4	0	3.5
4	23	7.5
4	50	10.2
4	78	.
4	106	18
4	134	.
4	162	25

Here is the thing, what I need to calculate is the average weight gained between each measure, that means between ages, as follows.

For example, Dog number 2 at age 0, or at birth, showed the weight of 3.7 kg and the next weight was performed at day 23 being 7.7 kg, it means that in 23 days this dog gained 4 kilograms, and it means also that each day the dog gained:

4/23 = 0.17391, it means 173.91 grams gained each day growing. As result of multiplication of this result in grams times 1000, converting this value from kilograms to grams.

With the information in table 1, we need to obtain the 3 last columns as shown in the table 2:

TABLE 2. RAW OBSERVED DATA, PLUS LAST 2 COLUMNS WITH KILOGRAMS AND GRAMS GAINED BY DAY

DOG	AGE_IN_DAYS	WEIGHT	days_between_measure	kg_gained	gr_gained_by_day
1	0	3.5	.	.	.
1	19	4.8	19	1.30	68.42
1	47	11.3	28	6.50	232.14
1	75	22	28	10.7	382.14
1	103	21.5	28	-0.5	-17.86
1	131	34.8	28	13.30	475.00
1	159	32.3	28	-2.50	-89.29
1	190	37.1	31	4.80	154.84
1	215	42.4	25	5.30	212.00
2	0	3.7	.	.	.
2	23	7.7	23	4.00	173.91
2	51	14.3	28	6.60	235.71
2	79	18	28	3.70	132.14
2	107	23.2	28	5.20	185.71
2	135	25	28	1.80	64.29
2	163	29.3	28	4.30	153.57
2	194	33.9	31	4.60	148.39
2	219	37.6	25	3.70	148.00
2	247	.	.	.	.
2	276	.	.	.	.
2	303	43.2	84	5.60	66.67
2	331	43	28	-0.20	-7.14
2	359	43.5	28	0.50	17.86
2	387	.	.	.	.
2	415	.	.	.	.
3	0	4.9	.	.	.
3	23	6.1	23	1.20	52.17
3	50	13.2	27	7.10	262.96
3	78	17.2	28	4.00	142.86
3	106	20	28	2.80	100.00
3	134	25	28	5.00	178.57
3	162	28	28	3.00	107.14
3	184	29	22	1.00	45.45
3	217	32.8	33	3.80	115.15
3	246	34.8	29	2.00	68.97
3	274	37.4	28	2.60	92.86
3	302	36.4	28	-1.00	-35.71
3	330	38.6	28	2.20	78.57
3	358	40.4	28	1.80	64.29
3	386	37.8	28	-2.60	-92.86
3	414	37.4	28	-0.40	-14.29
3	442	37.2	28	-0.20	-7.14
3	470	35.4	28	-1.80	-64.29
3	498	36.1	28	0.70	25.00
3	526	42	28	5.90	210.71
3	554	40.4	28	-1.60	-57.14
3	786	64.4	232	24.00	103.45
3	814	65.9	28	1.50	53.57
4	0	3.5	.	.	.
4	23	7.5	23	4.00	173.91
4	50	10.2	27	2.70	100.00
4	78	.	.
4	106	18	56	7.80	139.29
4	134	.	.
4	162	25	56	7.00	125.00

From here, what have to be obtained is the corrected weight each 20 days of live for each dog, i'll try to explain this in the next paragraphs.

If you look carefully each measure, at its corresponding line, has its own kilogram gained and grams gained by day. This value could be negative, it does not matter cause dogs often loose weight, this is pretty normal. Therefore, the question is I need to obtain the adjusted weight each 20 days from birth to further ages: 20, 40, 60, 80 and so on until the dog dies and has no more measures. This happens when the next information is at age 0 corresponding to the next dog.

For a better representation I will use dog 3 and its measures at day 50 and 78.

Dog 3 at 50 days old weighted 13.2 Kg, then at 78 days old he weighted 17.2. It means that in 28 days this do gained 4 kilograms, one more time as follows:

4/28 = 0.14285, it means 142.85 grams gained each day growing from 50 to 78 days old.

The question is: If I have these, and their corresponding daily gains weights, I can estimate other measures in other ages, for example each 20 days. I wanted to point out these measures because if I want to estimate the weight at 60 days old of the dog I can use this information of the 50 and 78 days old, having the estimate gain in grams each day I can estimate the weight at 60 days like this:

[(weight gain in grams between 2 measures)*(age to estimate weight-first age measured)] + weight at first age measured

[(142.85 grams)*(60-50)]+weight at 50 days = [(142.85)*10]+13.2 = 1.4285+13.2 = 14.6285 kg at 60 days.

From now, and using the table above obtained, we need to obtain the corrected weight at corresponding each 20 days, as follows in the next table:

DOG	AGE_IN_DAYS	WEIGHT	days_between_measure	kg_gained	gr_gained_by_day	new_age	weight_at_new_age
1	0	3.5	.	.	.	.
1	19	4.8	19	1.30	68.42	20	5.03214
1	47	11.3	28	6.50	232.14	40	9.67500
1	75	22	28	10.70	382.14	60	16.26786
1	103	21.5	28	-0.50	-17.86	80	21.91071
1	131	34.8	28	13.30	475.00	100	21.55357
1	159	32.3	28	-2.50	-89.29	120	29.57500
1	190	37.1	31	4.80	154.84	140	33.99643
1	215	42.4	25	5.30	212.00	160	32.45484
1	243	.	28	.	.	180	35.55161
1	272	.	29	.	.	200	39.22000
1	299	.	27	.	.	220	.
1	327	.	28	.	.	240	.
2	0	3.7	.	.	.	20	7.17826
2	23	7.7	23	4.00	173.91	40	11.70714
2	51	14.3	28	6.60	235.71	60	15.48929
2	79	18	28	3.70	132.14	80	18.18571
2	107	23.2	28	5.20	185.71	100	21.90000
2	135	25	28	1.80	64.29	120	24.03571
2	163	29.3	28	4.30	153.57	140	25.76786
2	194	33.9	31	4.60	148.39	160	33.00968
2	219	37.6	25	3.70	148.00	180	36.41600
2	247	.	28	.	.	200	34.78800
2	276	.	29	.	.	220	37.74800
2	303	43.2	27	.	.	240	40.70800
2	331	43	28	-0.20	-7.14	260	43.66800
2	359	43.5	28	0.50	17.86	280	46.62800
2	387	.	28	.	.	300	49.58800
2	415	.	28	.	.	320	43.07857
2	.	.	.	.	.	340	43.16071
3	0	4.9	.	.	.	20	5.94348
3	23	6.1	23	1.20	52.17	40	10.57037
3	50	13.2	27	7.10	262.96	60	14.62857
3	78	17.2	28	4.00	142.86	80	17.40000
3	106	20	28	2.80	100.00	100	19.40000
3	134	25	28	5.00	178.57	120	22.50000
3	162	28	28	3.00	107.14	140	25.64286
3	184	29	22	1.00	45.45	160	27.78571
3	217	32.8	33	3.80	115.15	180	28.81818
3	246	34.8	29	2.00	68.97	200	30.84242
3	274	37.4	28	2.60	92.86	220	33.00690
3	302	36.4	28	-1.00	-35.71	240	34.38621
3	330	38.6	28	2.20	78.57	260	36.10000
3	358	40.4	28	1.80	64.29	280	37.18571
3	386	37.8	28	-2.60	-92.86	300	36.47143
3	414	37.4	28	-0.40	-14.29	320	37.81429
3	442	37.2	28	-0.20	-7.14	340	39.24286
3	470	35.4	28	-1.80	-64.29	360	40.21429
3	498	36.1	28	0.70	25.00	380	38.35714
3	526	42	28	5.90	210.71	400	37.60000
3	554	40.4	28	-1.60	-57.14	420	37.35714
3	786	64.4	232	24.00	103.45	440	37.21429
3	814	65.9	28	1.50	53.57	460	37.32857
3	.	.	.	.	.	480	35.65000
3	.	.	.	.	.	500	36.52143
3	.	.	.	.	.	520	40.73571
3	.	.	.	.	.	540	41.20000
3	.	.	.	.	.	560	41.02069
3	.	.	.	.	.	780	63.77931
3	.	.	.	.	.	800	65.15000
4	0	3.5	.	.	.	20	6.97826
4	23	7.5	23	4.00	173.91	40	9.20000
4	50	10.2	27	2.70	100.00	60	11.59286
4	78	.	.			80	14.37857
4	106	18	56	7.80	139.29	100	17.16429
4	134	.	.			120	19.75000
4	162	25	56	7.00	125.00	140	22.25000

With the prerogative, IT IS NEEDED TO ESTIMATE THE WEIGHT OF EACH DOG EACH 20 DAYS. 

Each example i post here, i try to make it very useful to be helped easily, so the colors in the table 3, what is thought to be achieved, are going to help in the estimations, because those are special cases because the data of each dog is special. Here we go:

As said repetitively, each 20 days must appear the new weight for each dog. As seen in Table 3, dog 1 has its first recorded weight at 19 days old, it means that this weight can not be used to estimate the weight at 20 days old of this dog, meanwhile for dog 2 and 3 they have their respective first weight at day 23, for that reason it can be estimated the weight at day 20, and this measure is allocated at the same line that the weight at age 0, indicating that dogs with its second weight less than 20 days old will be its first estimated weight at 20 days in the second line, and those with the second weight with more than 20 days can have the new estimated weight at 20 days in the same line that the weight at age 0.

Else, for dog 1 (for weigh at 19 and 47) and for dog 3 (for weigh at 78 and 47) it can be observed that the new estimated measures at days 20 and 40 for dog 1, and at days 80 and 100 for dog 3, are used to estimate two new weights, using just 2 measures.

Next, in green color it can be observed that the new estimated measures are not located in the same line that the weights used to their respective estimation. It can be noted that these two estimations are result of their respective upper and lower ages. Describing this with details:

In dog 1 for estimated weights at 80 and 100 days of age, its used the recorded weight at 75 and 103 days, and those new estimates are located not in the same line that the corresponding line, those values are located in new rows below the last new estimated values, for this i will take this to see it here:

DOG	AGE_IN_DAYS	WEIGHT	days_between_measure	kg_gained	gr_gained_by_day	new_age	weight_at_new_age
1	75	22	28	10.70	382.14	60	16.26786
1	103	21.5	28	-0.50	-17.86	80	21.91071
1	131	34.8	28	13.30	475.00	100	21.55357

This can be observed also for the dog 2, for the estimated values for 140 and 160 ages, as seen in the TABLE 3:

DOG	AGE_IN_DAYS	WEIGHT	days_between_measure	kg_gained	gr_gained_by_day	new_age	weight_at_new_age
2	135	25	28	1.80	64.29	120	24.03571
2	163	29.3	28	4.30	153.57	140	25.76786
2	194	33.9	31	4.60	148.39	160	33.00968

Related case can be observed for dog 3 and for estimation of the new weight at 220 and 240 ages because the new data are in rows after the original data used to their estimation:

DOG	AGE_IN_DAYS	WEIGHT	days_between_measure	kg_gained	gr_gained_by_day	new_age	weight_at_new_age
3	217	32.8	33	3.80	115.15	180	28.81818
3	246	34.8	29	2.00	68.97	200	30.84242
3	274	37.4	28	2.60	92.86	220	33.00690
3	302	36.4	28	-1.00	-35.71	240	34.38621

Special cases can be observed in Dog 2 and in Dog 4, where no weights measures were taken for 247 and 276 days for the first one and the day 78 and 134, for the dog 4. In both cases the used values to estimate the middle values were the weights at 219 and 303 days in dog 2 and weight at ages 50, 106 and 134 for dog 4.

In dog 2, can be observed that the difference between days 219 and 303 is 84 dais, in these period the dog gained 5.6 kilograms and 66,67 grams per day, and using this information were estimated weights at 220, 240, 260 280 and 300 days of age, as shown in Table 3:

DOG	AGE_IN_DAYS	WEIGHT	days_between_measure	kg_gained	gr_gained_by_day	new_age	weight_at_new_age
2	219	37.6	25	3.70	148.00	180	36.41600
2	247	.	28	.	.	200	34.78800
2	276	.	29	.	.	220	37.74800
2	303	43.2	27	.	.	240	40.70800
2	331	43	28	-0.20	-7.14	260	43.66800
2	359	43.5	28	0.50	17.86	280	46.62800
2	387	.	28	.	.	300	49.58800

The other special case is shown in dog 4 where the is many missing data, where

In this case, the dog was not measured in relation to it weight at 78 and 134 days of age, but the other measures are recorded. In this case to estimate weights for 60, 80 and 100 days were used the recorded weights at 50 and 106 days, witch difference was 56 days with an gained weight of 7.8 kilograms and 139.29 grams per day. In the same context, to estimate weight for 120, 140 and 160 days of age were user the weights at 106 and 162 years of age.

DOG	AGE_IN_DAYS	WEIGHT	days_between_measure	kg_gained	gr_gained_by_day	new_age	weight_at_new_age
4	0	3.5	.	.	.	20	6.97826
4	23	7.5	23	4.00	173.91	40	9.20000
4	50	10.2	27	2.70	100.00	60	11.59286
4	78	.	.			80	14.37857
4	106	18	56	7.80	139.29	100	17.16429
4	134	.	.			120	19.75000
4	162	25	56	7.00	125.00	140	22.25000

Finally, it can be pointed out that the observations for each dog ends where the value for age of the next dog is equal to zero, so i guess the loop to be applied for each animal. With this, the table to be obtained, again, is:

DOG	AGE_IN_DAYS	WEIGHT	days_between_measure	kg_gained	gr_gained_by_day	new_age	weight_at_new_age
1	0	3.5	.	.	.	.
1	19	4.8	19	1.30	68.42	20	5.03214
1	47	11.3	28	6.50	232.14	40	9.67500
1	75	22	28	10.70	382.14	60	16.26786
1	103	21.5	28	-0.50	-17.86	80	21.91071
1	131	34.8	28	13.30	475.00	100	21.55357
1	159	32.3	28	-2.50	-89.29	120	29.57500
1	190	37.1	31	4.80	154.84	140	33.99643
1	215	42.4	25	5.30	212.00	160	32.45484
1	243	.	28	.	.	180	35.55161
1	272	.	29	.	.	200	39.22000
1	299	.	27	.	.	220	.
1	327	.	28	.	.	240	.
2	0	3.7	.	.	.	20	7.17826
2	23	7.7	23	4.00	173.91	40	11.70714
2	51	14.3	28	6.60	235.71	60	15.48929
2	79	18	28	3.70	132.14	80	18.18571
2	107	23.2	28	5.20	185.71	100	21.90000
2	135	25	28	1.80	64.29	120	24.03571
2	163	29.3	28	4.30	153.57	140	25.76786
2	194	33.9	31	4.60	148.39	160	33.00968
2	219	37.6	25	3.70	148.00	180	36.41600
2	247	.	28	.	.	200	34.78800
2	276	.	29	.	.	220	37.74800
2	303	43.2	27	.	.	240	40.70800
2	331	43	28	-0.20	-7.14	260	43.66800
2	359	43.5	28	0.50	17.86	280	46.62800
2	387	.	28	.	.	300	49.58800
2	415	.	28	.	.	320	43.07857
2	.	.	.	.	.	340	43.16071
3	0	4.9	.	.	.	20	5.94348
3	23	6.1	23	1.20	52.17	40	10.57037
3	50	13.2	27	7.10	262.96	60	14.62857
3	78	17.2	28	4.00	142.86	80	17.40000
3	106	20	28	2.80	100.00	100	19.40000
3	134	25	28	5.00	178.57	120	22.50000
3	162	28	28	3.00	107.14	140	25.64286
3	184	29	22	1.00	45.45	160	27.78571
3	217	32.8	33	3.80	115.15	180	28.81818
3	246	34.8	29	2.00	68.97	200	30.84242
3	274	37.4	28	2.60	92.86	220	33.00690
3	302	36.4	28	-1.00	-35.71	240	34.38621
3	330	38.6	28	2.20	78.57	260	36.10000
3	358	40.4	28	1.80	64.29	280	37.18571
3	386	37.8	28	-2.60	-92.86	300	36.47143
3	414	37.4	28	-0.40	-14.29	320	37.81429
3	442	37.2	28	-0.20	-7.14	340	39.24286
3	470	35.4	28	-1.80	-64.29	360	40.21429
3	498	36.1	28	0.70	25.00	380	38.35714
3	526	42	28	5.90	210.71	400	37.60000
3	554	40.4	28	-1.60	-57.14	420	37.35714
3	786	64.4	232	24.00	103.45	440	37.21429
3	814	65.9	28	1.50	53.57	460	37.32857
3	.	.	.	.	.	480	35.65000
3	.	.	.	.	.	500	36.52143
3	.	.	.	.	.	520	40.73571
3	.	.	.	.	.	540	41.20000
3	.	.	.	.	.	560	41.02069
3	.	.	.	.	.	780	63.77931
3	.	.	.	.	.	800	65.15000
4	0	3.5	.	.	.	20	6.97826
4	23	7.5	23	4.00	173.91	40	9.20000
4	50	10.2	27	2.70	100.00	60	11.59286
4	78	.	.	.	.	80	14.37857
4	106	18	56	7.80	139.29	100	17.16429
4	134	.	.	.	.	120	19.75000
4	162	25	56	7.00	125.00	140	22.25000

After this, and transposing this table, the new data would be:

DOG	0	20	40	60	80	100	120	140	160	180	200	220	240	260	280	300	320	340	360	380	400	420	440	460	480	500	520	540	560	580	600	620	640	660	680	700	720	740	760	780	800
1	3.5000	5.0321	9.6750	16.2679	21.9107	21.5536	29.5750	33.9964	32.4548	35.5516	39.2200	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
2	3.7000	7.1783	11.7071	15.4893	18.1857	21.9000	24.0357	25.7679	33.0097	36.4160	34.7880	37.7480	40.7080	43.6680	46.6280	49.5880	43.0786	43.1607	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
3	4.9000	5.9435	10.5704	14.6286	17.4000	19.4000	22.5000	25.6429	27.7857	28.8182	30.8424	33.0069	34.3862	36.1000	37.1857	36.4714	37.8143	39.2429	40.2143	38.3571	37.6000	37.3571	37.2143	37.3286	35.6500	36.5214	40.7357	41.2000	41.0207	.	.	.	.	.	.	.	.	.	.	63.7793	65.1500
4	3.5000	6.9783	9.2000	11.5929	14.3786	17.1643	19.7500	22.2500	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.

where in the first column appears the dogs evaluated, and in the next columns are the new estimated weights each 20 days.

I hope to be helped.

Thank you very much 

[ballardw]

Excel files are not data sets.

Please provide such in the form of a data step.

What  you are requesting is called Interpolation. Here is one way:

data have;
   input dog age_in_days weight;
datalines;
1  0 3.5
1 19 4.8
1 47 11.3
1 75 22
;

data interpolated;
   set have;
   by dog;
   lage= lag(age_in_days);
   lwt = lag(weight);
   if first.dog then ;
   else do ;
      deltawt= (weight - lwt) /(age_in_days - lage);
      newweight= lwt;
      do newage=lage to age_in_days;
         if mod(newage,20)=0 then output;
         newweight = newweight + deltawt;
      end;
   end;
   keep newweight newage;
run;

This actually calculates a weight each day and does have an end point /start point issue that will be masked for anything except an actual age multiple of 20 in the data. Exercise for the interested reader to remove it.

[jonatan_velarde]

Thank you very much for the answer:

as requested, here is the data set:

data have;

input DOG AGE_IN_DAYS WEIGHT;

cards;
1 0 3.5
1 19 4.8
1 47 11.3
1 75 22
1 103 21.5
1 131 34.8
1 159 32.3
1 190 37.1
1 215 42.4
1 243 .
1 272 .
1 299 .
1 327 .
2 0 3.7
2 23 7.7
2 51 14.3
2 79 18
2 107 23.2
2 135 25
2 163 29.3
2 194 33.9
2 219 37.6
2 247 .
2 276 .
2 303 43.2
2 331 43
2 359 43.5
2 387 .
2 415 .
3 0 4.9
3 23 6.1
3 50 13.2
3 78 17.2
3 106 20
3 134 25
3 162 28
3 184 29
3 217 32.8
3 246 34.8
3 274 37.4
3 302 36.4
3 330 38.6
3 358 40.4
3 386 37.8
3 414 37.4
3 442 37.2
3 470 35.4
3 498 36.1
3 526 42
3 554 40.4
3 786 64.4
3 814 65.9
4 0 3.5
4 23 7.5
4 50 10.2
4 78 .
4 106 18
4 134 .
4 162 25

;

[mkeintz]

Edit note: This program has been modified to:

1. Read in only DOG, AGE_IN_DAYS, and WEIGHT variables into dataset HAVE.
2. Fix the problem of extra observations in the NEW_AGES data set due to extra loop invoked when WEIGHT=.  The new statement now is (the bold italics is new):  

   if nxt_dog=dog and weight^=. then do new_age= 20*ceil(age_in_days/20) to nxt_age by 20;

Instead of creating 1 table with both weight gain history and weight at new ages, why not create two datasets:

1. weight_gain_history
2. new_ages

Now I understand all the new age weight estimates will be for 20, 40, 60, ... days, so the new_weight values are interpolated between the nearest preceding and nearest following recorded ages.  If so, then :

data have;
 input DOG	AGE_IN_DAYS	WEIGHT	;
datalines;
1	  0	 3.5
1    19	 4.8
1	 47	11.3
1	 75	22
1	103	21.5
1	131	34.8
1	159	32.3
1	190	37.1
1	215	42.4
2	  0	 3.7
2	 23	 7.7
2	 51	14.3
2	 79	18
2	107	23.2
2	135	25
2	163	29.3
2	194	33.9
2	219	37.6
2	247	  .
2	276   .
2	303	43.2
2	331	43
2	359	43.5
2	387	  .
2	415	  .
3	  0	 4.9
3	 23	 6.1
3	 50	13.2
3	 78	17.2
3	106	20
3	134	25
3	162	28
3	184	29
3	217	32.8
3	246	34.8
3	274	37.4
3	302	36.4
3	330	38.6
3	358	40.4
3	386	37.8
3	414	37.4
3	442	37.2
3	470	35.4
3	498	36.1
3	526	42
3	554	40.4
3	786	64.4
3	814	65.9
4	  0	 3.5	
4	 23	 7.5	
4	 50	10.2
4	 78	  .
4	106	18
4	134	  .	
4	162	25
run;

data weight_history  (drop=nxt_: new_:)
     new_ages (keep=dog new_:);
  set have;
  by dog;
  if weight^=. then do;
    days_between_measure=dif(age_in_days);
    kg_gained=round(dif(weight),.1);
	if first.dog then call missing(days_between_measure,kg_gained);
	if kg_gained^=. then gr_gained_by_day=  round(1000*kg_gained/days_between_measure,.01);
  end;
  output weight_history;
  if weight^=. then do until (nxt_weight^=.);
    set have (firstobs=2 keep=dog age_in_days weight rename=(dog=nxt_dog age_in_days=nxt_age weight=nxt_weight));
  end;
  if nxt_dog=dog and weight^=. then do new_age= 20*ceil(age_in_days/20) to nxt_age by 20;    
    new_weight=  round(weight + (nxt_weight-weight)*((new_age-age_in_days)/(nxt_age-age_in_days)),.00001);
    output new_ages;
  end;
run;

notes:

1.   The DIF function is defined as  dif(x)=x-lag(x).  But "lag" is a misnomer.  It really means "the most recent update in a queue of values of X.   So when you have "if weight^=. then kg_gained=dif(weight)", the "lag" value is only updated when non-missing values of weight are in the record.  I.e. it is never contaminated with missing weight values - so when a record follows a missing weight, then the "lag" value goes back 2 records, not 1.
2. Notice you can declare multiple output datasets in the DATA statement, and you can selectively output them.
3. The "if first.dog then call missing ..." statement ensures that you don't contaminate the starting value of one dog with the ending value of the preceding dog.
4. The "if weight^=. then do until (nxt_weight^=.) reads ahead one record until a non-missing "next" weight is encountered.
5. The following loop generates, through interpolation of WEIGHT ad NXT_WEIGHT, all the weights for intervening occurrences of 20, 40, 60, ... age in days.

Regards,

Mark

[jonatan_velarde]

Very nice code, but before this:

How do i create the column "days_between_measur"?

thanks in advance

[mkeintz]

It's right there in the code.  I suggest a cup of coffee and a second look.

Regards,

Mark

[jonatan_velarde]

Dog number 4 show like that:

76	4	0	3.5000
77	4	20	6.9783
78	4	40	9.2000
79	4	60	11.5929
80	4	80	14.3786
81	4	100	17.1643
82	4	80	.
83	4	100	.
84	4	120	19.7500
85	4	140	22.2500
86	4	160	24.7500
87	4	140	.
88	4	160	.

Could it be the loop?

Thanks in advance