Web & Social Media Analytics Previous Year Question Paper.pdf
CBMR notes
1. Chapter 16: Chi Square Analysis:Steps Page 375 measures proportion
When there is expected distribution like Small: Large: Family = 1:2:3
If significant reject null meaning NOT AS EXPECTED.
One Sample Mean test (Chi Square Analysis):t-test checks means
One sample mean test checks if sample mean is equal to population mean.
E.g. 10 store sample, Q is mean sales is 100 or not for all stores..
Steps on page 385.
If significant reject null meaning NOT AS EXPECTED.
Two Independent Sample Mean tests (Chi Square Analysis):
Checks association between 2 categorical variables. Page 385 - 395
NOTE: Hypothesis is always tested on POPULATION using samples.
Chapter 17: Anova page 438
You are comparing two samples and their means.
Example in book is sweets being sold in Plastic / Metal container: Q is whether there is impact of type of container…whether
plastics or metal.
e.g. Is the average job performance score(Dependent)different for those who have got different forms of training (Independent)
One way Anova steps page 442 - 445 for interpretation
Two One way Anova pages 454
Steps page 458 Plots - horizontal (mono variable)/ - Separate line
Linene's test - equal variance: If not then Games Howell(< 0.5 is equal variance)
Partial ETA Square: Impact - Effect checked through/ - Partial ETA Square
Chapter 18: Regression
Dependent Variable and Independent Variables relationship to be determined.
Bivariate: one x one y PAGE 469
Multiple regression: one y multiple x
Example in book: Sales = y Then Advertisement x
PAUL IS DAMN SURE REGRESSION WILL COME
2
Similar treatment for significance and interpretation. Also important is the R value.
Multiple Regressions
Dependence of two variables on y (page 477)
Check model summary and Anova
Analyse, regression. Linear
Put Dependent and Independent variable
Stats ensuremodel fit, estimate and descriptive
To interpret refer page 479
R, R-square, Standard error, Significance value, Beta.
Chapter 19: Discriminant Analysis.IT IS ABOUT DISCRIMINATION.
Set of group, people with different parameters…you try to differentiate based on the parameters.
Page 489 example is given;Page 497 Model is Significant from Anova
On Page 496 Canonical Discriminant Function Coefficients.
Chapter 20: Factor Analysis: Page 514.
Factors can be reduced to fewer only if they are highly correlated. Funda is if two or more factors are perfectly correlated we
can count just one, hence we can reduce the factors from many to few.
See steps…Component matrix and screen plot show answer
Interdependence of responses.
e.g. Why people use Facebook ? Check if there is any interdependence of responses?
Steps - page 516, Interdependence 521
KMO > 0.50, Bartlett's test, Rotated matrix - name in one group.
Chapter 21: Cluster Analysis:page 543 onwards.
Interdependence of objects is studied through Cluster Analysis and forming homogenous groups of cities.
Segment Identification, Used for finding groups within the data provided.
Important - 560 - 567: read 559, Inference Interpretation 564, Page 562 you can specify no. of Cluster,
Page 567 analyse from columns.