Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
1
00:00:01,320 --> 00:00:03,110
If the project is important,
2
2
00:00:03,110 --> 00:00:04,960
and you want to know what resources
3
3
00:00:04,960 --> 00:00:07,870
and material and services the deliverable needs,
4
4
00:00:07,870 --> 00:00:10,470
the activity breakdown is by far the best technique.
5
5
00:00:11,650 --> 00:00:15,060
However, if you just want a quick estimate,
6
6
00:00:15,060 --> 00:00:18,480
the Program Evaluation Review Technique, or PERT,
7
7
00:00:18,480 --> 00:00:22,550
is fast and a lot better than a single-point guess.
8
8
00:00:22,550 --> 00:00:25,493
As a rule of thumb, about plus minus 50% accuracy.
9
9
00:00:27,360 --> 00:00:30,100
The development of PERT was driven by the observation
10
10
00:00:30,100 --> 00:00:33,240
that most people produce low estimates
11
11
00:00:33,240 --> 00:00:36,350
because they are afraid of looking incompetent.
12
12
00:00:36,350 --> 00:00:39,003
They are afraid to tell you the real number.
13
13
00:00:40,400 --> 00:00:42,940
So the key value of this technique is
14
14
00:00:42,940 --> 00:00:44,513
it factors in the worst-case.
15
15
00:00:45,990 --> 00:00:48,500
The method estimates three numbers,
16
16
00:00:48,500 --> 00:00:50,460
the optimistic case O,
17
17
00:00:50,460 --> 00:00:52,500
most likely case M,
18
18
00:00:52,500 --> 00:00:55,370
and pessimistic case P,
19
19
00:00:55,370 --> 00:00:58,163
and then calculates the statistical likelihood.
20
20
00:00:59,020 --> 00:01:02,020
As the PERT estimate is equal to the optimistic case,
21
21
00:01:02,020 --> 00:01:03,860
plus four times the most likely case,
22
22
00:01:03,860 --> 00:01:06,843
plus the pessimistic case, all divided by six.
23
23
00:01:08,390 --> 00:01:11,920
The most likely estimate M is multiplied by four
24
24
00:01:11,920 --> 00:01:13,700
to approximate data if it was drawn from
25
25
00:01:13,700 --> 00:01:16,040
a normal statistical distribution,
26
26
00:01:16,040 --> 00:01:18,340
which is the usual distribution of most events,
27
27
00:01:18,340 --> 00:01:20,010
sometimes called the Bell Curve,
28
28
00:01:20,010 --> 00:01:23,030
where numbers are more likely to fall in the middle.
29
29
00:01:23,030 --> 00:01:25,823
So to get the average you have to divide by six.
30
30
00:01:27,360 --> 00:01:29,660
Now here's the value of the technique.
31
31
00:01:29,660 --> 00:01:31,930
After trying to look competent by estimating
32
32
00:01:31,930 --> 00:01:35,110
a smaller number for the optimistic and most likely case,
33
33
00:01:35,110 --> 00:01:38,130
people almost always estimate the pessimistic estimate
34
34
00:01:38,130 --> 00:01:41,310
as much larger than the most likely number.
35
35
00:01:41,310 --> 00:01:44,440
So the overall PERT average provides a larger
36
36
00:01:44,440 --> 00:01:46,723
and therefore more realistic estimate.
37
37
00:01:47,840 --> 00:01:49,690
We'll look at an example in a minute.
38
38
00:01:52,000 --> 00:01:55,100
Interestingly however, humans being human,
39
39
00:01:55,100 --> 00:01:56,870
experience shows the PERT estimate
40
40
00:01:56,870 --> 00:01:59,590
often still underestimates in practice,
41
41
00:01:59,590 --> 00:02:02,350
so a straight 3-point average,
42
42
00:02:02,350 --> 00:02:04,940
with no weighting of the most likely number by four,
43
43
00:02:04,940 --> 00:02:06,523
is often recommended today.
44
44
00:02:07,430 --> 00:02:09,610
The 3-point estimate is equal to the optimistic,
45
45
00:02:09,610 --> 00:02:12,893
plus most likely, plus pessimistic, all divided by three.
46
46
00:02:14,610 --> 00:02:16,750
If you're going to use one of the two methods,
47
47
00:02:16,750 --> 00:02:19,580
straight 3-point is recommended instead of PERT,
48
48
00:02:19,580 --> 00:02:21,240
since underestimation is a much more
49
49
00:02:21,240 --> 00:02:23,393
common problem than overestimation.
50
50
00:02:25,530 --> 00:02:27,720
If you want to make really sure the estimate
51
51
00:02:27,720 --> 00:02:29,960
is large enough you can also pad it
52
52
00:02:29,960 --> 00:02:32,500
with one or more standard deviations,
53
53
00:02:32,500 --> 00:02:34,860
calculated as the standard deviation is equal to the
54
54
00:02:34,860 --> 00:02:38,283
optimistic minus the pessimistic estimate divided by six.
55
55
00:02:39,750 --> 00:02:41,640
Then statistical theory tells us
56
56
00:02:41,640 --> 00:02:43,700
that if you add one standard deviation
57
57
00:02:43,700 --> 00:02:46,220
the actual result should be less than or equal to
58
58
00:02:46,220 --> 00:02:48,483
your estimate 84% of the time.
59
59
00:02:49,760 --> 00:02:51,810
If you add two standard deviations,
60
60
00:02:51,810 --> 00:02:53,520
the actual result should be less than or equal
61
61
00:02:53,520 --> 00:02:56,023
your estimate 98% of the time.
62
62
00:02:56,900 --> 00:02:59,130
And if you add three standard deviations,
63
63
00:02:59,130 --> 00:03:00,810
the actual result should be less than or equal
64
64
00:03:00,810 --> 00:03:03,743
the estimate 99.9% of the time.
65
65
00:03:05,650 --> 00:03:07,313
Let's take a look at an example.
66
66
00:03:09,580 --> 00:03:11,910
Let's say a deliverable owner estimates that their work
67
67
00:03:11,910 --> 00:03:14,880
will most likely take 15 days to complete,
68
68
00:03:14,880 --> 00:03:17,220
could take 10 days if all goes perfectly,
69
69
00:03:17,220 --> 00:03:19,060
or could take as much as 30 days
70
70
00:03:19,060 --> 00:03:20,853
if there are significant problems.
71
71
00:03:22,560 --> 00:03:25,640
Then the PERT estimate is 10 plus four
72
72
00:03:25,640 --> 00:03:30,640
times 15 plus 30 all divided by six or 16.7.
73
73
00:03:31,290 --> 00:03:33,820
A little bit more than 10% higher
74
74
00:03:33,820 --> 00:03:35,320
than the most likely estimate.
75
75
00:03:37,000 --> 00:03:40,930
The 3-point estimate is 10 plus 15 plus 30
76
76
00:03:40,930 --> 00:03:45,070
straight divided by three or 18.3.
77
77
00:03:45,070 --> 00:03:47,300
A little more than 20% higher
78
78
00:03:47,300 --> 00:03:48,800
than the most likely estimate.
79
79
00:03:50,300 --> 00:03:53,290
If you have a choice, the 3-point estimate is recommended,
80
80
00:03:53,290 --> 00:03:54,980
since, as is usually the case,
81
81
00:03:54,980 --> 00:03:58,380
it is larger than PERT, and in the real world,
82
82
00:03:58,380 --> 00:03:59,783
likely to be more accurate.
83
83
00:04:02,000 --> 00:04:03,800
If you want to add some padding,
84
84
00:04:03,800 --> 00:04:07,080
the standard deviation is SD equals 30 minus 10
85
85
00:04:07,080 --> 00:04:09,350
divided by six or 3.3.
86
86
00:04:09,350 --> 00:04:11,890
So padding the PERT estimate indicates that,
87
87
00:04:11,890 --> 00:04:14,720
if this estimate is drawn from a normal distribution,
88
88
00:04:14,720 --> 00:04:18,300
adding one standard deviation means that 84% of the time
89
89
00:04:18,300 --> 00:04:21,743
16.7 plus 3.3 or 20 days should be enough.
90
90
00:04:23,080 --> 00:04:26,190
Adding two standard deviations means that 98% of the time
91
91
00:04:26,190 --> 00:04:31,003
16.7 plus 3.3 plus 3.3 or 23.3 days should be enough.
92
92
00:04:32,010 --> 00:04:33,990
And adding three standard deviations mean that
93
93
00:04:33,990 --> 00:04:38,250
99.9% of the time 16.7 plus 3.3 plus 3.3 plus 3.3
94
94
00:04:38,250 --> 00:04:40,903
or 26.6 days should be enough.
95
95
00:04:42,710 --> 00:04:44,730
You can also use statistical formulas
96
96
00:04:44,730 --> 00:04:47,640
to determine how many standard deviations to add
97
97
00:04:47,640 --> 00:04:50,450
to obtain any given percentage confidence you want.
98
98
00:04:50,450 --> 00:04:53,863
For example, 60%, 70%, 80%, et cetera.
99
99
00:04:56,240 --> 00:04:58,560
There's no question, PERT and 3-point
100
100
00:04:58,560 --> 00:05:01,790
are faster techniques than activity breakdown,
101
101
00:05:01,790 --> 00:05:05,220
however they are just not as accurate.
102
102
00:05:05,220 --> 00:05:07,340
And the basic reason is that
103
103
00:05:07,340 --> 00:05:09,620
the larger the item being estimated,
104
104
00:05:09,620 --> 00:05:13,070
the less accurate any estimating technique will be,
105
105
00:05:13,070 --> 00:05:16,730
and with PERT and 3-point we typically want to move fast,
106
106
00:05:16,730 --> 00:05:20,520
ao by definition are estimating larger items of work,
107
107
00:05:20,520 --> 00:05:21,843
not at the activity level.
108
108
00:05:23,130 --> 00:05:25,510
Which returns us to the fundamental point.
109
109
00:05:25,510 --> 00:05:29,240
If the project and an accurate estimate are important,
110
110
00:05:29,240 --> 00:05:31,130
instead work with the core project team
111
111
00:05:31,130 --> 00:05:33,720
to break the effort into individual deliverables,
112
112
00:05:33,720 --> 00:05:34,970
and then have the leads break
113
113
00:05:34,970 --> 00:05:37,123
the deliverables into activities.
114
114
00:05:38,300 --> 00:05:41,380
And then you get all the other advantages as well,
115
115
00:05:41,380 --> 00:05:44,060
such as identification of the resources,
116
116
00:05:44,060 --> 00:05:46,423
material, and services required.
10237
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.