Statistics

Front 1

What are the two types of data you can encounter?

Back 1

1) Categorical

2) Continous

Front 2

What's referred as categorical data?

Back 2

Data points consist of distinct categories.

No inherent order or numerical meaning.

Front 3

What's referred as continous data?

Back 3

Measurements of any numerical value within a range.

Front 4

What are the measurements for categorical data? (4)

Back 4

1) Frequency count

2) Mode

3) Proportion

4) Cross-tabulation (contingency table)

Front 5

What's cross-tabulation?

Back 5

Also known as contingency table puts in relation two categorical variables.

One variable forming the rows (e.g. male/female) and one forming the columns (e.g. employed/unemployed).

Front 6

What are the measurements for continuous data? (4)

Back 6

1) Central Tendency

2) Spread

3) Shape (Skewness)

4) Correlation and Regression

Front 7

What are the measures of Central Tendency? (3)

Back 7

1) Mean

2) Median

3) Mode

Front 8

What are the measures of Spread? (5)

Back 8

1) Range

2) Interquartile Range (IQR)

3) Variance

4) Standard Deviation

5) Coefficient of Variation

Front 9

What are the measures of Shape? (2)

Back 9

1) Skewness (symmetry)

2) Kurtosis (flatness)

Front 10

Moments are statistical ___ used to describe ___ of a ___ distribution.

Why the term moment?

Back 10

Measurements, characteristics, probability

A moment of a mathematical function describe its shape and behavior.

Front 11

Enumerate the moments of statistics (4)

Back 11

1) Central Tendency: Mean (mu)

2) Spread: Variance (sigma squared)

3) Symmetry: Skewness (gamma)

4) Flatness: Kurtosis (kappa)

Front 12

Null hypothesis (H0) claims that there is NO significant ___ between ___, or NO ___ between ___.

The word "null" stands for the ___ of ___.

Back 12

Correlation, variables, effect, treatments

Absence, relationship

Front 13

The p-value is the ___ of the observed data to ___ if the null hypothesis was ___.

You choose a ___ level (a) in advance if ___ < ___ then it's OK to ___ the null hypothesis.

Back 13

Probability, occure, true

Significance, p < a, reject

Front 14

Correlation is the measure of ___ and ___ of a linear ___ between two ___.

Correlation doesn't imply ___; it simply indicates the ___ to which changes in one variable are ___ with changes in other one.

Back 14

Strentgh and direction

Relationship, variables

Causation

Degree, associated

Front 15

Regression is used to ___ the relationship between a ___ variable and one or more ___ variables.

Regression helps ___ the value of a ___ variable based on the values of ___ variables.

Back 15

Model, dependent, independent

Predict, dependent, independent

Front 16

What are the measures of correlation? (2)

Back 16

1. Correlation coefficients

2. Scatter plots

Front 17

What are the measures of regression? (5)

Back 17

1. Regression analysis

2. Coefficient of determination

3. Model evaluation

4. Residual analysis

5. Significance testing

Front 18

Make an example of Regression Analysis

Back 18

Regression analysis can help you build a mathematical model that, for instance, relates years of experience (independent variable) to salary (dependent variable) and make predictions for new individuals.

E.g. Salary = 1000 + 50 * Years / 4

Front 19

Make an example of Coefficient of Determination

Back 19

After performing a regression analysis, you calculate an R-squared value of let's say 0.85.

This means that 85% of the variability in the dependent variable can be explained by independent variable, indicating a strong correlation between the two.

Front 20

Make an example of Residual Analysis

Back 20

Residual analysis involves plotting the differences between the predicted prices and actual prices to check for patterns or trends. If you notice a pattern in the residuals, it may indicate that your model has systematic errors that need to be addressed.

Front 21

Make an example of Significance Testing

Back 21

You perform a significance test, such as a t-test or ANOVA, to determine if there is a statistically significant difference between the treatment and control groups.

This helps you determine whether the drug has a real effect or if the results could have occurred by chance.

Front 22

Make an example of Model Evaluation

Back 22

You have trained a model to classify emails as spam or not. To evaluate its performance, you use a dataset of 1,000 emails with known labels. You measure its accuracy, precision, recall, and F1-score to assess how well it classifies emails correctly.