실리콘밸리의 비개발자 직장인

*EWMA (Exponentially Weighted Moving Average) / 지수가중이동평균 : To exponentially forget the past data, we want to attach more weight to the most recent data 과거의 데이터를 빨리 잊어버리기 위해 현재의 데이터에 더 많은 가중치를 부여한다. Zi = monitoring statistics , Xi = observation, λ=current, (1-λ) = history weight when λ=1 , => Shewhart chart * UCL / LCL of EWMA * In general : Var(Zt) = (σ^2/n)(λ/(2-λ))(1-(1-λ)^2t) * Steady state c..

* Process Capability Analysis : 제조 공정을 안정 상태로 유지한 경우 그 공정이 만들어내는 품질의 달성 능력을 말한다. [네이버 지식백과] 공정 능력 [process capability, 工程能力, こうていのうりょく] (용어해설) - Cp, Cpk, Cpkm - C.I and Hypothesis testing for Cp * Process Capability Ration : Cp = distance between specification limits / 6*sigma - 프로세스가 양측 규격인 경우 적용 - Cp = 1 means the process are uses up 100% tolerance band with 0.27% (2700ppm) nonconforming units..

* Cost - benefit analysis (비용편익분석) - estimate the strengths and weakness of alternatives that satisfy transcations, activities or functional requirement for the business 투자재원의 한계 내에서 어떤 사업을 선택하는것이 바람직한 것인지 분석 - benefits : labor, time and cost-savings - purpose: 1) to determine if it is a sound sound investment/ decision (justification/ feasibility) 2) to provide basis for comparing benefits and ..

S chart 는 언제 쓰면 될까 ? 1) Sample size 'n > 10' 일 때 ( n이 10보다 작거나 같을 경우 R 차트를 사용하는것이 좋다) 2) Used for variable sample size => Subgroups have different sample size 1. X bar and S control chart with variable sample size (subgroup의 sample size가 다를 경우) m : # of samples/ subgroups ni : sample size for each samples/ subgroups * 각 subgroup의 sample size가 다르지만 , 하나의 고정된 컨트롤 리밋으로 작성하고 싶을 경우에는 ni의 평균을 사용하거나 가장..

컨트롤 차트 , X_bar and R​ - Suppose (xij, i=1..m j=1..n} are normally distributed - X bar chart monitors between-sample variability (variability over time) X bar 차트는 샘플들 간의 변동성을 나타내고 R chart measures within-sample variablitiy (intantaneous variability at a given time) R 차트는 샘플내의 변동성에 대해 측정한다. 1. Control chart for X_bar and R - Known μ and σ - Range Ri = max (x_ij) - min (x_ij) for j=1.. n - μ and σ을..

* Root and Cause Analysis , Continuous Improvement - To prevent recurrence at lowest cost in the simplest way 발본색원의 개념, 근원이 되는 문제를 제거해서 비슷한 문제들이 다시는 발생하지 않도록 방지하는 것 - method to trace down RC: data mining hierachial cluster solution (GT data mining) 1) Root cause: removed; prevent undesireable outcome,the condition that enable one or more causes 2) Casual cause: affect an event's outcome, but not..

- Dissimilarity Matrix (n objects x p attributes) : Store a collection of proximities that are available for all pairs of n objects A triangular matix, i.e., d(i,j) = d(j,i) d(i,j) is the measured dissimilarity or "differce" between objects i and j similarity measure sim(i,j) = 1 - d(i,j) d(i,j) is a non-negative number that is close to 0 when objects i and j are highly similar or "near" each ot..

- Quartiles : Q1 (25th percentile), Q3 (75th percentile) - Inter-quartile range: IQR = Q3- Q1 - Five number summary : min, Q1, median, Q3, max - Boxplot : ends of the box are the quartiles; median is marked ; add whiskers (minimum and maximum observation) ​ and plot outliers individually ; whiskers : 박스 바깥쪽에 가로로 나있는 선, 최소값과 최대값을 표시 - Outlier : usually, a value higer/ lower than 1.5 x IQR ex) 30,..

- Data set 은 data objects로 이루어져 있고 data object는 entity 를 표현한다. ex) sales database: customers, store items, sales medical database: patients, treatments - Data objects는 attributes에 의해 설명된다. - Database의 row는 data objects로 columns는 attributes를 나타낸다. - Attribute 의 종류에는 1) Nominal 2) Binary 3) Ordinal 4) Numeric: quantitative 가 있다. 1) Nominal : categories, states or symbols - Hair_color = {auburn, bl..

* Measure of dependent/correlated events: lift 1) Statistiacal independence P(S∩B) = P(S) x P(B) => Statistically independence P(S∩B) > P(S) x P(B) => Postively correlated P(S∩B) Negatively correlated 2) lift : interestingness measure: (interesting = reasonable 정도로 이해하면 될 것 같다) - if the lift is equal to 1, then A and B are independent and there is no correlation between them. A⊥..