ANOVA – Analysis of Variance (Test of means)

  • ANOVA model – Basic principle is “A dependent continuous variable is a function of one or more categorical independent variables. Each categorical variable has two or more levels”
  • Variables can be 1 or 2 or n. Accordingly ANOVA model would be One-Way, Two-Way or n-Way(or MANOVA – Multiple ANOVA)
  • SCENARIO: Productivity based on type of drinks consumed by employees on the floor. The number shows the calls per day taken by given employee consuming the given drink for the day. 
    • This is a One-Way ANOVA with Beverages as variable and Juice, Coffee and Tea as Factors/Levels
    • NULL Hypothesis, H0: Mean of Juice = Mean of Coffee = Mean of Tea
    • Alternate Hypothesis, Ha: At least one mean is different
  • Total Variance of the data = Variation between groups + Variation within groups
  • F-Statistic = Variation between groups / Variation within groups
  • NOTE: F-Statistic is always a positive value and the distribution curve is always a positive curve
    • F-Statistic = 0 is shown as a starting point in the normal curve
    • F-Statistic can have values from 0 till infinity
    • When Variation b/w Groups is high, then Variation within groups will be less or vice versa


REFERENCES

Comments