1.
Probability Distribution
1. A probability distribution describes how probabilities are assigned to each possible value of a
random variable.
2. It is of two types: discrete (binomial, Poisson) and continuous (normal, exponential).
3. Every probability lies between 0 and 1, and total probability of all outcomes equals 1.
4. It helps model uncertainty in engineering, manufacturing, quality control, etc.
5. Mean, variance, and standard deviation are important characteristics of a distribution.
6. Graphically represented by PMF (for discrete) or PDF (for continuous).
---
2. Parametric vs Non-Parametric Tests
1. Parametric tests assume the population is normally distributed and parameters (mean, SD) are
known.
2. Examples: t-test, z-test, ANOVA, F-test.
3. Suitable for numerical and continuous data with larger sample sizes.
4. Non-parametric tests do not require any distribution assumption.
5. Examples: Chi-square test, Mann–Whitney test, Kruskal–Wallis test.
6. Best for ordinal, ranked, or small-sample data.
---
3. Type-I & Type-II Errors
1. Type-I error (α): Rejecting a true null hypothesis.
2. Example: Concluding a good machine is defective.
3. Type-II error (β): Accepting a false null hypothesis.
4. Example: Concluding a defective part is acceptable.
5. α represents level of significance; β relates to power of the test.
6. Reducing α increases β, so both must be balanced.
---
4. Level of Significance & Confidence Level
1. Level of significance (α) is the probability of Type-I error.
2. Common α values: 5%, 1%.
3. Lower α means stricter decision-making in testing.
4. Confidence level = 1 – α. (e.g., α = 0.05 → 95% confidence)
5. Higher confidence gives wider intervals; lower confidence gives narrower intervals.
6. Used in hypothesis testing and confidence interval estimation.
---
5. Mean, Variance & Probability Distribution
1. Mean: Average value of data.
2. Shows central tendency of a dataset.
3. Variance: Measures spread of data around the mean.
4. High variance → high variability.
5. Probability distribution gives likelihood of different outcomes.
6. Examples: Normal distribution (continuous), binomial distribution (discrete).
---
6. Null Hypothesis, Alternative Hypothesis & Test of Significance
1. Null hypothesis (H■) assumes no difference/effect.
2. Alternative hypothesis (H■) indicates a difference/effect exists.
3. Test of significance checks whether sample data supports rejecting H■.
4. Uses test statistics like z, t, chi-square.
5. Decision made based on p-value or critical value.
6. Example: Testing if mean life of bulbs is equal to 1000 hours.
---
7. Karl Pearson’s Correlation Method
1. Measures strength and direction of linear relationship between two variables.
2. Correlation coefficient (r) ranges from –1 to +1.
3. r ≈ +1: strong positive, r ≈ –1: strong negative, r ≈ 0: no relation.
4. Formula uses covariance and standard deviation of variables.
5. Helps understand relation before performing regression.
6. Widely used in engineering, QC, forecasting.
---
8. Procedure of Regression Analysis
1. Identify dependent variable (Y) and independent variable (X).
2. Assume linear model Y = a + bX.
3. Use least squares method to calculate constants a and b.
4. Develop the regression equation.
5. Use the equation to predict future or unknown values.
6. Check fit using correlation and residuals.
---
9. Steps in Correlation & Regression Analysis
1. Collect paired data (X, Y).
2. Draw a scatter diagram to observe pattern.
3. Calculate means of X and Y.
4. Compute correlation coefficient r.
5. Calculate regression coefficients a, b.
6. Form regression lines and interpret slope.
---
10. Any One Sampling Type (Simple Random Sampling)
1. Every member gets equal chance of selection.
2. Eliminates personal bias.
3. Can be done using lottery or random numbers.
4. Best for homogeneous population.
5. Simple to use and understand.
6. Example: Randomly selecting 20 items from 500 manufactured items.
---
11. Sampling Techniques
1. Random sampling: Equal chance for all units.
2. Stratified sampling: Divide population into groups and sample from each.
3. Systematic sampling: Select every k■■ item.
4. Cluster sampling: Select entire groups/clusters randomly.
5. Multistage sampling: Multiple levels of sampling.
6. Reduces cost, time, and improves representation.
---
12. Sampling Distribution & Central Limit Theorem
1. Sampling distribution is the distribution of a sample statistic (mean, proportion).
2. Used to estimate parameters of a population.
3. CLT states: For large samples (n > 30), sample mean follows normal distribution.
4. Works even if population distribution is not normal.
5. Mean of sampling distribution = population mean.
6. Standard error SE = σ / √n.
---
13. Procedure of Hypothesis Testing
1. State H■ and H■ clearly.
2. Select significance level α
3. Choose appropriate test (z, t, chi-square).
4. Compute test statistic.
5. Compare with table value or p-value.
6. Conclude to reject or fail to reject H■.
---
14. Curve Fitting by Least Squares Method
1. Least squares minimizes the sum of squared deviations between observed and fitted values.
2. Choose form of curve (straight line, parabola, exponential).
3. Form normal equations for unknown constants.
4. Solve equations to get parameters.
5. Substitute values to get fitted curve.
6. Used for predicting trends and interpolation.
---
15. Fitting a Straight-Line Trend
1. Trend equation: Y = a + bX.
2. Actual mean method: Uses deviations from actual means of X and Y.
3. Assumed mean method: Uses a convenient assumed mean to simplify calculations.
4. Regression coefficients computed using least squares formula.
5. Best-fit straight line represents long-term trend.
6. Used in forecasting, production planning, time series analysis.
---
16. Steps in One-Way ANOVA
1. State H■: All group means are equal.
2. Calculate group means and overall mean.
3. Compute Sum of Squares Between (SSB) and Within (SSW).
4. Calculate Mean Square values: MSB = SSB/(k−1), MSW = SSW/(N−k).
5. Compute F-ratio = MSB / MSW.
6. Compare with F-table and conclude.