Application of Probability in textile manufacturing
Dr. BAPI SAHA
Dept. of Mathematics
Govt. College of Engg. & Textile Technology
Berhampore
Areas to be covered:
• Basic concepts of probability
• Conditional probability
• Baye’s theorem
• Some discrete and continuous distributions
Classical definition of probability:
• The probability of an event is given by
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑎𝑣𝑜𝑟𝑎𝑏𝑙𝑒 𝑐𝑎𝑠𝑒𝑠
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑐𝑎𝑠𝑒𝑠
• Example: Coin tossing of an unbiased coin.
• What if the coin is biased?
• Drawback of classical definition: It is applicable only when the outcomes are equally likely.
• Axiomatic definition:
1. 0 ≤ 𝑃(𝐴) ≤ 1
2. 𝑃 𝜑 = 0
3. 𝑃 =𝑖𝑛ڂ1 𝐴𝑖 = σ𝑛𝑖=1 𝑃 𝐴𝑖 when 𝐴′𝑖 s are disjoint events.
Conditional probability:
• Let us try to figure out the probabilities of the following two events:
1. 𝑃 𝑟𝑎𝑖𝑛 𝑓𝑎𝑙𝑙 𝑖𝑛 𝑎 𝑝𝑎𝑟𝑡𝑖𝑐𝑢𝑙𝑎𝑟 𝑑𝑎𝑦 𝑜𝑓 𝑎𝑛 𝑦𝑒𝑎𝑟
2. 𝑃 𝑟𝑎𝑖𝑛 𝑓𝑎𝑙𝑙 𝑖𝑛 𝑎 𝑝𝑎𝑟𝑡𝑖𝑐𝑢𝑙𝑎𝑟 𝑑𝑎𝑦 𝑜𝑓 𝑎𝑛 𝑦𝑒𝑎𝑟 𝑔𝑖𝑣𝑒𝑛 𝑖𝑡 𝑖𝑠 𝑡ℎ𝑒 𝑚𝑜𝑛𝑡ℎ 𝑜𝑓 𝐽𝑢𝑙𝑦
It is obvious that𝑷 𝒓𝒂𝒊𝒏 𝒇𝒂𝒍𝒍 𝒊𝒏 𝒂 𝒑𝒂𝒓𝒕𝒊𝒄𝒖𝒍𝒂𝒓 𝒅𝒂𝒚 𝒐𝒇 𝒂𝒏 𝒚𝒆𝒂𝒓 𝒈𝒊𝒗𝒆𝒏 𝒊𝒕 𝒊𝒔 𝒕𝒉𝒆 𝒎𝒐𝒏𝒕𝒉 𝒐𝒇 𝑱𝒖𝒍𝒚 ≥
𝑷 𝒓𝒂𝒊𝒏 𝒇𝒂𝒍𝒍 𝒊𝒏 𝒂 𝒑𝒂𝒓𝒕𝒊𝒄𝒖𝒍𝒂𝒓 𝒅𝒂𝒚 𝒐𝒇 𝒂𝒏 𝒚𝒆𝒂𝒓
Sampling and its distribution
• Sample: A sample is a collection of independent and identically distributed (i. i. d) random variables.
• Example of sample in textile technology
• Different sampling techniques:
1. Simple Random Sampling: In this sampling technique, each member of the population is equally likely to be included in
the sample.
a. Simple random sampling with replacement (SRSWR): In this case the selected item is returned back before the
next draw.
Total possible number of samples of size 𝑛 is 𝑁 𝑛
b. Simple random sampling with out replacement (SRSWOR): In this case the selected item is not returned back before the
next draw.
𝑁
Total possible number of sample of size n is 𝑛
Statistics
• Sample mean: Let 𝑋1 , 𝑋2 , ⋯ , 𝑋𝑛 be a sample of size 𝑛. Then sample mean is
σ 𝑛
𝑋
• 𝑋ത = 𝑖=1 𝑖
𝑛
σ𝑛 ത 2
𝑖=1 𝑋𝑖 −𝑋
• Sample variance: 𝑆2 =
𝑛−1
• Statistic: A statistic is a function of sample.
• Example: Sample mean, sample variance etc.
Distribution of sample mean:
Theorem: If 𝑋1 , 𝑋2 , ⋯ , 𝑋𝑛 constitute a random sample of size 𝑛 from an infinite population having mean and
𝜎 2
variance 𝜇 and 𝜎 2 respectively, then 𝐸 𝑋ത = 𝜇 and 𝑣𝑎𝑟 𝑋ത = 𝑛
Central Limit Theorem: If 𝑋1 , 𝑋2 , ⋯ , 𝑋𝑛 constitute a random sample of size 𝑛 from an infinite population with
ത
𝑋−𝜇
mean 𝜇, variance 𝜎 2 , and moment generating function 𝑀𝑋 𝑡 , then 𝑍 = follows standard normal
𝜎/ 𝑛
distribution as 𝑛 → ∞.
Contd…
• Chi-square Distribution
A continuous random variable 𝑋 is said to follow chi-square distribution if its probability density function is
𝑥 𝜈
− −1
𝑒 2𝑥2
given by 𝑓 𝑥 = 𝜈
𝜈
; 𝑥 > 0, 𝜈 > 0
22 𝛤 2
= 0; elsewhere
How is it related to normal distribution?
Theorem:
If 𝑋 has the standard normal distribution, then 𝑋 2 has the 𝜒 2 distribution with 𝜈 = 1 degree of freedom.
Acceptance sampling:
• In attribute sampling plan, a sample of size 𝑛 is taken at random from a batch and if the number of
defectives found in the sample is greater than a given acceptance number 𝑐, the batch is rejected.
• In acceptance sampling suitable number of sample size 𝑛 and a number 𝑐 is set such that if out of 𝑛
number of items, the number of defectives are less than or equal to 𝑐, the sample is accepted.
• If p is the proportion of defectives in the lot, the probability that a sample of size n will be accepted
is given by
𝑃𝑎 𝑝 = 𝑃 𝑟 ≤ 𝑐
= 𝑃 𝑟 = 0 + 𝑃 𝑟 = 1 + ⋯+ 𝑃 𝑟 = 𝑐
𝑛 𝑛 𝑛
= 0
𝑝0 1 − 𝑝 𝑛−0
+ 1
𝑝1 1 − 𝑝 𝑛−1
+ ⋯+ 𝑐
𝑝𝑐 1 − 𝑝 𝑛−𝑐
𝑛
= σ𝑐𝑘=0 𝑘
𝑝𝑘 1 − 𝑝 𝑛−𝑘
Operating Characteristic curve (OC curve) and its significance
• The plot of acceptance probability vs the proportion of defective (p) is known as OC curve.
Acceptable Quality Level( 𝐴𝑄𝐿)(𝑝1 )=limiting
proportion of defective such that if the proportion of
defective lies below 𝑝1 , the sample is accepted.
Lot tolerance proportion defective (LTPD)(𝑝2 ) =
limiting proportion of defective such that if the
proportion of defective exceeds 𝑝2 , the sample is
rejected.
Producer’s risk: The sample is rejected though the
actual proportion of defective lies below 𝑝1 ; 𝛼 = 1 −
𝑝𝑎 𝑝1
Consumer’s risk: The sample is accepted though the
actual proportion of defective exceeds 𝑝2 ; 𝛽 =
𝑝𝑎 𝑝2
OC curve contd…
• In ideal case, the probability of acceptance of a lot
should be 1 as long as the proportion of defective remains below
𝑝1 .
Determining 𝑛 and 𝑐
𝑛
• 𝑃𝑎 𝑝1 = 1 − 𝛼 = σ𝑐𝑘=0 𝑘
𝑝1𝑘 1 − 𝑝1 𝑛−𝑘
𝑐 𝑛 𝑘 𝑛−𝑘
• 𝑃𝑎 𝑝2 = 𝛽 = σ𝑘=0 𝑘 𝑝2 1 − 𝑝2
These two equations are highly non-linear and is not easy to solve as 𝑛 and 𝑐 are integers. The
problem is overcome by using Chi-square distribution. The three governing equations reduces to
𝜒 21−𝛼,2 𝑐+1 = 2𝑛𝑝1
𝜒 2𝛽,2 𝑐+1 = 2𝑛𝑝2
2
𝜒𝛽,2 𝑐+1 𝑝2
2 =
𝜒1−𝛼,2 𝑐+1 𝑝1
Thank you
