s7 Fundamental Steps to Complete a Data Analytics Project

• Step 1: Understand the Business. ...

• Step 2: Get Your Data. ...
• Step 3: Explore and Clean Your Data. ...
• Step 4: Enrich Your Dataset. ...
• Step 5: Build Helpful Visualizations. ...
• Step 6: Get Predictive. ...
• Step 7: Iterate, Iterate, Iterate.

EXPLORATORY DATA ANALYSIS(EDA)

Introduction to EDA:
• Exploratory data analysis refers to the critical process of performing initial investigation on data

so as to discover patterns, to spot anomalies, to test hypothesis and to check assumption with
the help of summary and graphical representation.

• It is a good practice to understand the data fast and try to gather as many insights from it.

Importance of EDA:
• Identifying the most important variables/features in your dataset.

• Testing a hypothesis or checking assumption related to dataset.

• To check the quality of data for further processing and cleaning.

• Deliver data driven insights to business stakeholders.

• Verify expected relationships actually exists in the data.

• To find unexpected structure or insights in the data.

Architecture of EDA:
• Business Objective

• Data Requirement

• Data Collection

• Exploratory Data Analysis

 • Modeling

• Evaluation

• Deployment

• Monitoring

Data Science Process:

• Ask interesting and related questions.

• Get the data.

• Explore the data.

• Model the data.

• Communicate and visualize the data.

Types of Data:
Structured data:

Csv file, Excel file, Database file

Unstructured data:

Text, images, videos, audios.

Major Types of Data:

Numerical
• Represent some sort of quantitative data measurement

• Discrete Data: Integer based; often counts of some data

• Continuous Data: Has an infinite number of possible value

Categorical
• Quantitative data that has no inherent mathematical meaning

• You can assign numbers to categories in order to represent them more compactly, but the
numbers don’t have mathematical meaning.

Ordinal
 • Simple numbering such as : rating and low/high/average

Types of Graph in Python Library:

• Bar Graph: Shows relation between different proportions

• Pie Chart: Represents percentage/proportions

• Histogram: Consider range

• Scatter Plot: Present plot by the help of dots

• Heat Map: Shows the count of different categories

• Box Plot: Works under the control of range

• Line Plot: Connects all point by forming a line to represent

• Violin Plot: Shows the maximum value and minimum value

DATA PRE-PROCESSING
 Data Preprocessing:
Definition: Data preprocessing involves cleaning and organizing raw data to make it suitable for
analysis or model training.
Tasks: Handling missing values, removing duplicates, dealing with outliers, and scaling numerical
features are common data preprocessing tasks.
Purpose: The goal is to ensure the dataset is consistent, accurate, and free from errors, enabling
effective analysis or model training.

1) Getting Dataset
2) Importing Libraries
3) Importing Datasets
4) Finding Missing Values
5) Encoding Categorical Data
6) Splitting Dataset into Training and Test set
7) Feature Scaling

When you get Data first of all take these four steps and see the:
1) Distribution of Distribution of Data
2) Composition of variables
3) Relationship of each variable with each other
4) Comparison between the variables
 DATA WRANGLING
 Data wrangling is the process of cleaning, transforming, and organizing data so, that it can be
used for analysis and visualization. It is an important part of data analysis because raw data is
often incomplete, inconsistent, and in an unstructured format, which makes it difficult to work
with.
 Data wrangling helps to make the data more consistent, accurate, and useful for data analysis
and decision-making.
 Data preprocessing, data wrangling, and feature engineering are three essential steps in the
data preparation process for machine learning and data analysis. While they are interconnected,
each serves a distinct purpose.
 Data Wrangling:
Definition: Data wrangling is the process of transforming and mapping raw data from its initial
format into another format that is more appropriate for analysis or modeling.
Tasks: Merging datasets, reshaping data, and handling categorical variables are typical data
wrangling tasks.
Purpose: Data wrangling aims to make the data more structured and suitable for downstream
analysis, facilitating easier exploration and modeling.
 In summary, data preprocessing focuses on cleaning and preparing raw data, data wrangling
involves transforming data into a more usable format, and feature engineering aims to enhance
the predictive power of models by creating or modifying features. These steps are integral to the
overall data preparation process, contributing to the success of machine learning and data
analysis tasks.

Steps to Clean Data:

1) Gathering data
- Import Data
- Load Data
- Read Data
2) Tools to clean data
- Libraries
3) How to clean data
 Assessing Data (EDA)
 Cleaning Data
- Dealing with missing values
- Correcting errors in the data (Outliers removal)
1) Visualization (Use to find outliers)
2) IQR Method (Use to remove outliers)
3) Z-score
- Dropping duplicates (Removing the entry which is taken more than one time)

Visualization and IQR are enough for data wrangling, the Z-score is an additional way.

Steps to Transform Data:

In data transformation, we do Data Normalization (Normalize the Data).

- Min-Max Normalization
- Standard Scaler
- Log Transformation
- Winsorization
- Z-score Normalization
- Decimal Scaling

Min Max Scaler scales the data to a fixed range, typically between 0 and 1.

Standard Scaler standardizes a feature by subtracting the mean and then scaling to unit variance. Unit
variance means dividing all the values by the standard deviation. Standard Scaler does not meet the
strict definition of scale I introduced earlier. (Value subtracted by mean and divide standard deviation (x
– mean(x) ) / standard deviation(x) ).

Log Transformation is the most popular among the different types of transformations used to transform
skewed data to conform to normality. If the original data follows a log-normal distribution or
approximately so, then the log-transformed data follows a normal or near-normal distribution. It does
not have a fixed range it makes a range according to the data.

Min Max Scaler, Standard Scaler, and Log Transformation are enough left one are additional ways.

Steps to Organize Data:

Organizing the data

 Columns Creation
 Renaming Columns (With a sensible name)

Save the Data

Feature Engineering:
 Feature Engineering:
Definition: Feature engineering involves creating new features or transforming existing ones to
enhance the performance of machine learning models.
Tasks: Creating interaction terms, transforming variables, and encoding categorical features are
common feature engineering tasks.
Purpose: Feature engineering helps models capture relevant patterns and relationships in the
data, leading to improved model accuracy and generalization.

 represents a measurable piece of data that can be used for analysis

 Features usually have a lot of data
 based on features machine learning models are trained

The feature is also known as:

  Input
 X
 Predictors

Steps in Feature Engineering:

1) Feature Cleaning (Features should be right)

- Anomaly
- Missing Values
- Outliers
 Know the Brainstorm
- Composition
- Distribution
- Relationship
- Comparison
2) Feature Creation (Creating a new feature)
3) Ml Model Testing (How do these contribute to each model training (Ranking of Features))
4) Re-iteration \Iterative Recreation (creating new features on the basis of ML model)
5) Feature Selection

Steps after Getting Dataset:

 Domain Knowledge
- Meta Data
- Full form of columns name (Abbreviation)
- Numerical Units
- Layout of Feature Engineering Process
 Data-Preprocessing
- Handling Missing Data
- Outliers detection and handling
 Numerical Features
- How many are there
- Convert numeric into numeric such as 10000 vs 1,0000
- Cleaning and spiliting
- Normalization or scaling the data
- Handling Skewness
- Binning
 Categorical Features
- Types of categorical features (Nominal or Ordinal)
- Encoding based on the category type (Label encoding\targeted encoding\one hot encoding)
(Conversion of categorical variables to numbers)
 Date and Time feature
- Reformat
- Spiliting time and dates\ convert data into on unit
- Binning (summer time and winter time)
- Count the time in one unit Such as total minutes in an hour
- Cyclic time count
 - Should know all the days in months such as leap, 30, 31 days
 Text Features
- Natural Language Programming NLP basics
1) Tokenize
2) Vectorize
3) TF-IDF
4) Bag of Words
5) Words embedding
6) Topic Modelling
 Image Features
- Computer Vision
- Transerflow Library
 Feature Selection
- Filter Method
1)
- Wrapper Method
- Embeded Methods
- Dimensionality Reduction
 Advanced Feature Engineering Technique
 - Regularization Method
 Eep learnig Based methood

Introduction to Statistics
 Branch of mathematics used to collect, analyze, interpret and make conclusions from data.
 The practice or science of collecting and analyzing numerical data in large quantities,
especially for the purpose of inferring proportions in a whole from those in a representative
sample.
 Methods for collecting, analyzing, interpreting and presenting empirical data.
 Data --> Statistics --> Information
 Two fundamental ideas of statistics are:
1) Uncertainty
2) Variation

Data:
Facts, especially numerical facts, collected together for reference or information.

Information:
Knowledge communicated concerning some particular fact.

Dataset:
A collection of related sets of information that is composed of separate elements but can
be manipulated as a unit by a computer.

Types of Dataset:
 Primary
- User Generated
- Too costly
- Time consuming
- Selfmade Data
 Secondary
- Use another’s generated data
- Cheap and sometimes free
- Not so time consuming
- Readymade data

Major Parts of Statistics:

1) Descriptive Statistics
2) Inferential Statistics

Descriptive Statistics:
- Presenting, organizing and summarizing data
- Describe the targeted population
- Organize, analyze and present the data in meaningful manner
- Result shown in charts, graphs, and tables
- Describe any data

Descriptive Analysis:
 Graphical
- Arrange data in tables
- Bar Graphs and Pie Charts
 Numerical
- Percentages
- Averages
- Range
 Relationships
- Correlation coefficient
- Regression Analysis

Inferential Statistics:
- Drawing conclusions about a population based on data observed in a sample.
- Make inferences from population and generalize the population
- Compares, test and predict futures upcoming
 - Result shown in probability scores
- Used to make conclusions and prediction on the available data

Inferential Analysis:
 Confidence interval
 Margin of Error
 Compare means of two samples
- Pre\Post Scores
- t –Test
 Compare means from three samples
- Pre\post and follow up
- ANOVA = analysis of variance

Three stages of Statistics:

 Design\Planning
 Descriptive Statistics
 Inferential Statistics
Phase I :Designing and Planning

 Purpose
- If data is used to explore and analyze so, it is descriptive
- If data is used to make results, conclusion or predictions so, it is inferential.

 Population vs Sample
- Census vs Survey

 Methodology
- Select the Variable
- Organize the Data variable

 Scale\Levels of Measurement
Start from
- Step1: Nominal
- Step2: Ordinal
- Step3: Interval
- Step4: Ratio
End to

 Data Types\Types of Variables

- Categorical
1) Nominal
2) Ordinal
 3) Qualitative
4) No numerical representation
5) Represented in texts
- Numerical
1) Continuous (Float)
2) Discrete (Integer)
3) Quantitative
4) Numerical
5) Mostly represented in Number

 Types of Variable on the Basis of their Nature:

- Dependent vs Independent
- Time Series vs Cross Sectional Data

 Types of Studies
- Observational (Only Observes) (Descriptive Analysis)
1) Can only claim association not causation.
2) Observes individuals only no attempt to influence outcome
- Experimental (Consideration, Inference) (Inferential Analysis)
1) Can claim causation
2) Applied a treatment to individuals to isolate it’s effect

Sampling Techniques:
 Probability
- Simple random sampling.
- Systematic sampling.
- Stratified sampling.
- Cluster sampling.
- Multistage Sampling

 Non Probability
- Convenience sampling.
- Judgmental\Purposive sampling.
- Snowball sampling.
- Quota sampling.

Probability:
Probability is simply how likely something is to happen. Whenever we're unsure about the outcome of
an event, we can talk about the probabilities of certain outcomes—how likely they are.

It is measured on 0 to 1 scale.

P(A) = m\n
m = number of ways that are favorable to the occurrence of A

n = total number of the outcomes of the experiment

Non Probability:
Non-probability sampling is a method of selecting units from a population using a subjective (i.e. non-
random) method.

Independent Events:
Two events A and B are said to be independent if the occurrence of A is in no way influenced by
the occurrence of B. Likewise occurrence of B is in no way influenced by the occurrence of A

What is the difference between probability and non-probability?

Probability Sampling is a sampling technique in which samples taken from a larger population are
chosen based on probability theory. Non-probability sampling method is a technique in which the
researcher chooses samples based on subjective judgment, preferably random selection

1. Simple Random Sampling:

Selecting individuals from a population in a purely random manner, where each member has an
equal chance of being chosen. Example: Drawing names from a hat to select participants for a
survey.

2. Systematic Sampling:
Choosing every nth individual from a list, with a fixed interval, after selecting a random starting
point. Example: Surveying every 10th customer in a store's loyalty program list.

3. Stratified Sampling:
Dividing the population into subgroups (strata) based on certain characteristics and then
randomly sampling from each stratum. Example: Selecting students for a study by first dividing
them into grade levels and then randomly selecting from each grade.

4. Cluster Sampling:
Dividing the population into clusters or groups and randomly selecting some of these clusters,
then surveying all members within the selected clusters. Example: Sampling a few
neighborhoods in a city and surveying all households in those neighborhoods.

5. Multistage sampling:
Multistage sampling is a complex sampling technique that involves selecting a sample from a
population in multiple stages or steps. It is often used when it is impractical or too expensive to
directly sample the entire population. In multistage sampling:
 The population is first divided into smaller, more manageable clusters or subpopulations.
 A random sample of these clusters is selected in the first stage.
 Within each selected cluster, further sampling stages may be conducted. This can involve
random sampling of individuals or units within the chosen clusters.

6. Convenience Sampling:
Choosing the easiest and most readily available participants, often leading to a non-
representative sample. Example: Surveying people who pass by a street corner.

7. Purposive\ Judgmental Sampling:

Selecting participants based on specific criteria or purpose, rather than random selection, often
used in qualitative research. Example: Interviewing experts in a particular field for a research
project.

8. Snowball Sampling:
Starting with one or a few participants and asking them to refer or recruit others for the study,
commonly used when the population is hard to reach. Example: Studying a rare disease by
asking diagnosed individuals to refer others with the same condition.

9. Quota Sampling:
Dividing the population into subgroups and setting a quota for each subgroup to ensure
proportional representation, but not through random selection. Example: Surveying a specific
number of males and females from different age groups in a market research study.

1. Nominal Scale:
Categorical data with no inherent order or ranking. Example: Colors (e.g., red, blue, green).

2. Ordinal Scale:
Categorical data with a specific order or ranking. Example: Education levels (e.g., high
school, bachelor's, master's).

3. Interval Scale:
Numerical data with equal intervals between values, but no true zero point. Example:
Temperature in Celsius (e.g., 20°C, 30°C).

4. Ratio Scale:
Numerical data with equal intervals and a true zero point, allowing for meaningful ratios.
Example: Height in centimeters (e.g., 160 cm, 180 cm).

Nominal vs Ordinal vs Interval vs Ratio:

Nominal data represents categories or labels with no inherent order, while ordinal data has
ordered categories but lacks equal intervals. Interval data has ordered categories with equal
 intervals, and ratio data has ordered categories with equal intervals and a meaningful zero
point.

Phase II :Descriptive Statistics or (EDA)

 Data Organization
- Tabulation
- Frequency Distribution (Repetition of any value in a variable is called frequency)
 Data Visualization
- Graphical Representation
- Shapes of Distribution
1) Bell-Distribution (Only Correct form )
2) Uniform
3) J- Shaped
4) Reverse J-Shaped
5) Right Skewed
6) Left Skewed
7) Bimodal
8) U-Shaped

 Data Summary
- Data is complex
- We need values to summarize it
- Central Tendency
- Dispersion

Types of Charts/Graphs
 Categorical Data
- Bar Charts
- Pie Charts

 Numerical Data
- Histogram
- Frequency Polygons
- OGIVE
Reference Must Read = Andrew Abela Plotting Guide

Event:
An event is an outcome of an experiment.

Experiment:
An experiment is a process that is performed to understand and observe possible outcomes.

Parameter:
A parameter is a number describing a whole population

Statistic:
A statistic is a number describing a sample.

Parameter Vs Statistic\Estimator:
The key point to understand is that a statistic is a measurable function of the sample and is a
random variable. Estimator: It is a statistic that does not depend on theta. An estimate of a parameter is
generally an unobservable quantity that describes the property of the underlying data distribution.
Quartile:
A type of quantile which divides the number of data points into four parts.

Interquartile:
In statistics, the interquartile range (IQR) is a measure of statistical dispersion that describes the spread
or variability of a dataset. It is based on quartiles, which are values that divide a data set into four equal
parts. The IQR specifically focuses on the middle 50% of the data, excluding the lower and upper 25%.

Here's the complete concept of the interquartile range:

- First Quartile (Q1): This is the value below which 25% of the data falls. It's the 25th
percentile.
- Second Quartile (Q2): This is the median of the data, which is the value that divides the data
into two equal halves. It's also the 50th percentile.
- Third Quartile (Q3): This is the value below which 75% of the data falls. It's the 75th
percentile.
- Fourth Quartile (Q4): This is the 100% value of the data falls. It's the 100th percentile.

Empirical Relationship:
In statistics, for a moderately skewed distribution, there exists a relation between mean, median and
mode. This mean median and mode relationship is known as the “empirical relationship” which is
defined as Mode is equal to the difference between 3 times the median and 2 times the mean.

Measure of Dispersion:
 Absolute Measures
- Range
- Quartile Deviation
- Mean Absolute Deviation
- Standard Deviation
  Relative Measures
- Coefficient of Variation
- Coefficient of Quartile Deviation
- Coefficient of Mean Deviation

Phase III :Inferential Statistics

 Parameter Estimation
- Changing Purpose of data from Descriptive to Inferential Statistics
- Interval Estimation
- Point Estimation
- Confidence Interval

 Hypothesis Testing
- Null Hypothesis
- Alternate Hypothesis

Parameter Estimation
The change in the response associated with a one-unit change of the predictor, all other
predictors being held constant.

Test Types:
To find difference between two value/variables etc. Two Test are used:

 t- Test
 z- Test

To find difference between Three or more value/variables etc. Test are used:

 ANOVA

Confidence Interval:
The probability that a population parameter will fall between a set of values for a certain proportion of
times.

Confidence Interval = (Point estimate) +or- (critical value) * (standard error)

s
CI= X ±z* √ n
CI = Confidence interval
X = sample mean
z = the chosen value
n= sample
s= standard deviation

If we have more data we would be more confident.

Hypothesis Testing Outline:
Normal Distribution:
Normal distribution with mean 0 and standard deviation is 1 is called a Normal Standard Deviation.
For samples of any size drawn from a normally distributed population, the sample mean is normally
distributed, with mean μX = μ and standard deviation σX=σ/√n, where n is the sample size.

 Associated by Bell curve and Mean, Median, Mode is used to know the central tendency.
 In this mean median and mode are divided equally
 It is symmetrical about its mean

Sampling Distribution:
A probability distribution of a statistic that is obtained through repeated sampling of a specific
population. It describes a range of possible outcomes for a statistic.
 Analyze the sample and make inferences about the population
 Sample statistic vs Population Parameter
 Sampling of a particular sample statistic of al possible samples that can be drawn from a
population
 Sampling distribution that mean

Central Limit Theorem:

if you take sufficiently large samples from a population, the samples' means will be normally
distributed, even if the population isn't normally distributed.
Properties:
 If an sample is drawn from a population that has a mean and a standard deviation
 Applies to both Continuous and Discrete
 Random Variable should have a well defined mean and variance (Standard deviation)
 Applicable even when the original variable is not normal.
 The sampling distribution follows a normal distribution with:
- Mean
- Standard Deviation
Assumptions:
 The data must be randomly sampled
 The samples values must be independent of each other
 The 10% condition: When the sample is drawn without replacement, the sample size n, should
be no more than 10% of the population

Inferential Statistics: Test and their Type

 Parametric
- More reliable result
- Meet the assumption
 Non Parametric
- Less reliable result
- Calculates the rank of the data
 - No need to meet assumptions

Steps to Complete Before Starting Data Analysis then use the Parametric or Non
Parametric :

1) Normality Test:
 Shapiro-Wilk Test
- Specific (Reliable)
- Data = normal
- Null Hypothesis is not acceptable if value of P is less than 0.5 (Data is not normal)
 Kolmogorov- Smirnov Test
- General (Less Reliable)
 Histplot, Histogram
 Boxplot
 Bell curve
 QQnom plot

2) Homogeneity Test
- The variance of the variable in data are equal
- Defines the distribution of data within its values by comparing
- Divide in group to compare data
 Levene’s Test
- Null Hypothesis is not acceptable if value of P is less than 0.5 (Data is not normal)

3) Purpose:
- Know the purpose of your Research Questions

 Types of Purpose:
 Comparison
- Check the difference
- Pearson’s Correlation
- Spearman’s Correlation
- At least two groups

 Relationship
- Check the connection
- Connection
- Correlation
- Causation
- Prediction
4) Data Type:
 - Know the type of the data on which we are working
 Categorical
- Nominal
- Ordinal
- Qualitative
- No numerical representation
- Represented in texts
- Yes or No, True or False
 Numerical
- Continuous (Float)
- Discrete (Integer)
- Quantitative
- Numerical
- Mostly represented in Number
- Amount, number, [Link]
5) Statistical Test:
 Choose a statistical test from three main families:
1) Chi- Squared
- Comparison
- Categorical only
2) t-Test \ ANOVA
- Comparison
- Categorical and Numerical
3) Correlation
- Relationship
- Numerical only
chi squared test:
 Chi squared test for homogeneity
 Chi squared test for Independence
When to use:
 Nothing effect this
 Can be used with any numbers of levels or groups

t- Test \ ANOVA:
 One sample t - test
 Two sample t- test
- Unpaired (Two different groups)
- Paired (Same group twice)
 ANOVA (Analysis of Variance)
- One- way ANOVA
- Two – way ANOVA
- Repeated Measure of ANOVA
Correlation:
 Person’s Correlation (One independent and one dependent)
- Tells us how closely two variables are connected.
 Regression (One independent and one dependent
- Tell us a specific mathematical equation that describes a relationship.

Assumptions about the Data:

 The data is normally distributed
 Follow a Gaussian Distribution
If assumptions are completed result would be non reliable.

If assumption do not meet:

1) Normalize your Data:
 Standardization
 Min max scale
 Log Transformation
2) Use alternative Non Parametric Test

Alternative Non Parametric test:

 Chi Squared = Chi Squared
 One sample t - test = one sample Wilcoxon signed rank test
 Two sample t- test
- Unpaired = Mann whiteney U -test
- Paired = Wicoxon
 ANOVA = Kruskal Wallis Test
 Correlation = Spearman’s Correlation & Kendall Tau
 Regression

Some other test:

 Reliability test
 Inter rater reliability test
 Validity Tests
 Sample size consumption

Hypothesis Testing:

A hypothesis is a claim or statement about a population.

A hypothesis test is a standard procedure for testing a claim about a property of population.
Key concepts:

 Null hypothesis
- Result is equal
- If want to reject any claim make it null hypothesis
- H ° should be defined or null hypothesis is rejected
 Alternative hypothesis
- Result is greater or less than the other value
 Test statistic
 Critical region
 Significant level
 Critical value
 P-value
 Type I and II error
1. Null hypothesis:
A statement that there is no significant difference or effect in a given experiment or study.

Example: In a drug trial, the null hypothesis might state that the new medication is not more effective
than a placebo in reducing blood pressure.

2. Alternative hypothesis:
A statement that contradicts the null hypothesis, suggesting there is a significant difference or
effect.

Example: In the same drug trial, the alternative hypothesis would state that the new medication is
more effective than a placebo in reducing blood pressure.

3. Test statistic:
A numerical value calculated from sample data used to assess whether the null hypothesis
should be accepted or rejected.

Example: In a t-test comparing the means of two groups, the test statistic is the t-value that measures
the difference between the group means.

4. Critical region:
The range of values in the distribution of a test statistic that, if the test statistic falls within,
leads to rejection of the null hypothesis.

Example: If the critical region for a significance level of 0.05 is t > 1.96 for a two-tailed t-test, and the
calculated test statistic is t = 2.50, the null hypothesis is rejected because the test statistic falls in the
critical region.

5. Significance level:
 The predetermined probability of making a Type I error (false positive) when testing a
hypothesis.

Example: A significance level of 0.05 means there is a 5% chance of concluding an effect exists when
there is no real effect (Type I error).

6. Critical value:
A specific value in the distribution of a test statistic that separates the critical region from the
non-critical region.

Example: In a z-test for a 95% confidence interval (α = 0.05), the critical value is 1.96 for a two-tailed
test.

7. P-value:
The probability of obtaining a test statistic as extreme as, or more extreme than, the observed
result under the null hypothesis.

Example: If the calculated P-value in a hypothesis test is 0.03, it indicates a 3% probability of obtaining
the observed result if the null hypothesis is true.

8. Type I and II error:

Errors in hypothesis testing - Type I error occurs when the null hypothesis is wrongly rejected,
and Type II error occurs when the null hypothesis is wrongly accepted.

Example: In a criminal trial, a Type I error would be convicting an innocent person (false positive),
while a Type II error would be acquitting a guilty person (false negative).

The null hypothesis (Denoted by H0 (Not) ) is a statement that the value of a population parameter
(such as proportion, mean, or standard deviation) is equal to some claimed value.

 We test the null hypothesis is directly proportional

 Either reject null hypothesis or fail to reject null hypothesis

Alternative Hypothesis (denoted by H1 or Ha or HA) is the statement that the parameter has a value that
somehow differs from the null hypothesis.

 The symbolic form of alternative hypothesis must use one of these symbols ≠, <, >.

IQR = Q3 - Q1. To detect the outliers using this method, we define a new range, let's call it decision
range, and any data point lying outside this range is considered as outlier and is accordingly dealt with.
The range is as given below: Lower Bound: (Q1 - 1.5 * IQR) Upper Bound: (Q3 + 1.5 * IQR)

Hypothesis Testing Outline:

Normal Distribution:
Normal distribution with mean 0 and standard deviation is 1 is called a Normal Standard
Deviation.
For samples of any size drawn from a normally distributed population, the sample mean is normally
distributed, with mean μX = μ and standard deviation σX=σ/√n, where n is the sample size.

 Associated by Bell curve and Mean, Median, Mode is used to know the central tendency.
 In this mean median and mode are divided equally
 It is symmetrical about its mean

Sampling Distribution:
A probability distribution of a statistic that is obtained through repeated sampling of a specific
population. It describes a range of possible outcomes for a statistic.
 Analyze the sample and make inferences about the population
 Sample statistic vs Population Parameter
 Sampling of a particular sample statistic of al possible samples that can be drawn from a
population
 Sampling distribution that mean

Central Limit Theorem:

if you take sufficiently large samples from a population, the samples' means will be normally
distributed, even if the population isn't normally distributed.
Properties:
 If an sample is drawn from a population that has a mean and a standard deviation
 Applies to both Continuous and Discrete
 Random Variable should have a well defined mean and variance (Standard deviation)
 Applicable even when the original variable is not normal.
 The sampling distribution follows a normal distribution with:
- Mean
- Standard Deviation
Assumptions:
 The data must be randomly sampled
 The samples values must be independent of each other
 The 10% condition: When the sample is drawn without replacement, the sample size n, should
be no more than 10% of the population

Other Important Terms:

1) Significance level:
The predetermined probability of making a Type I error (false positive) when testing a
hypothesis.
Significance level (denoted by α) is the probability that the test statistic will fall in the critical
region when the null hypothesis is actually true. Common choices for α are 0.05, 0.01, and 0.10.

Example: A significance level of 0.05 means there is a 5% chance of concluding an effect exists when
there is no real effect (Type I error).

2) P-value:
 The probability of obtaining a test statistic as extreme as, or more extreme than, the observed
result under the null hypothesis. Denoted by α.
The p value or P value (Probability Value) is the probability of getting a value of the test statistic
that is at least as extreme as the one representing the sample data, assuming that the null
hypothesis is true. The null hypothesis is rejected by if the P value is very small, such as 0.05 or
less.

Example: If the calculated P-value in a hypothesis test is 0.03, it indicates a 3% probability of obtaining
the observed result if the null hypothesis is true.

3) Critical region:
The range of values in the distribution of a test statistic that, if the test statistic falls within,
leads to rejection of the null hypothesis.
Critical region is the set of values of the test statistic that cause us to reject the null hypothesis

Example: If the critical region for a significance level of 0.05 is t > 1.96 for a two-tailed t-test, and the
calculated test statistic is t = 2.50, the null hypothesis is rejected because the test statistic falls in the
critical region.

4) Critical value:
A specific value in the distribution of a test statistic that separates the critical region from the
non-critical region.

Example: In a z-test for a 95% confidence interval (α = 0.05), the critical value is 1.96 for a two-tailed
test.

Conclusions
In Hypothesis Testing:
We always test the null hypothesis .

The initial conclusion will always be one of the following:

 Reject the null hypothesis

 Fail to reject the null hypothesis

Decision Criterion:

Traditional Method:

 Reject H(Not) if the test Statistic falls within the critical region.
 Fail to Reject H(Not) if the statistic does not fall within the critical region.
Random Variables:
 Random Experiment
 Random Variable
 Types of Random Variable
 Probability Distributions

Applying Tests For EDA:

t-test:

A t - test is a statistic that checks if two means are reliably different from each other.

t = variance between groups/ variance within groups

A big t = different groups

A small t value = similar groups

 Independent sample t test

Test the mean of different groups
 Paired sample t test
Test the mean of one group twice
 One Sample t test
Tests the mean of one group against a set mean.

Machine Learning
 Machine Learning allows machine to learn and make decisions smartly.
 Machine can learn from the data provided or their own experience. It depends on the of
Machine Learning.
 Data -> Learning -> Prediction -> Decision

Machine Learning Life Cycle :

1) Data Collection
 Identify various source of information
 Gather Data
 Combine the data acquired from various data source
2) Data Preparation
 Deals with the exploration of your data to generate for better result
  Understand as much as you can
3) Data wrangling
 Filtering \ Cleaning raw data
 Filtering Noise
 Recognizing and removing the outliers
 Removing or filling missing values
4) Data Modeling
 Selecting machine learning algorithms
 Building the models
 Validating the results
5) Model Training
 Training model is a process in which a machine learning algorithm is fed with sufficient
training data to learn from
6) Model Testing
 This involves checking for accuracy of the model by providing the with the input that are
seen
7) Deployment
 Model ready deployment for prediction
 Monitoring the mode as per requirement

Each step is very important for the accuracy of the model.

Types of Machine Learning:

 Supervised
- Works under supervision
- Somebody teaches
- Prediction
- Outcome
- Input Data -> Training model -> Prediction -> Output or Report
- Classification for Categories
- Regression for Numerical
 Unsupervised\Clustering
- No supervision
- No teacher
- Self learning
- No labelling of data
- Find a pattern by itself
- Training model -> Prediction -> Output or Report
 Semi Supervised
- Have both characteristics of Supervised and Unsupervised
- Some data is labelled some are not
- Input Data -> Training model \Clustering-> Prediction -> Output or Report
 Reinforcement
 - Hit and Trial learning
- Learn from mistakes
- Reward and Punishment Rule
- Prediction is based on reward or punishment
- Depends on feedback
- Learning Process: Input Data -> Training model
- Working Process: Prediction -> Output or Report

Supervised Machine Learning

In supervised learning, you train the machine learning using data which is well “labeled”, some input
data is already tagged with correct answer and this algorithm learns from labeled training data that
helps you to predict the further outcomes.

Step:1) Labeled Dataset

Step:2) Model Training

Step:3) Prediction

Supervised ML is further divided into:

1) Regression
2) Classification

Supervised ML Algorithms:

1) Regression
 Linear Regression
 Polynomial Regression
 Regression Trees
2) Classification
 Random Forest
 Decision Trees
 Logistic Regression
 Support Vector Machine

Difference between regression and classification is that in regression output is numeric while on other
hand classification output is categorical.

Advantages of Supervised ML:

 You have full control on what machine is learning

 You can easily train or debug model
 You can determine the number of classes

Disadvantages of Supervised ML:

 Have limited space

  Collecting labeled dataset is expensive and time consuming
 Wrong prediction

Unsupervised Machine Learning

In unsupervised learning, you train the machine using data which is “unlabeled” ,and model itself find
the hidden pattern and insights from the given data.

Step:1) Unlabeled Dataset

Step:2) Model \Interpretation Training

Step:3) Algorithm

Step:4) Processing

Step:5) Output

Unsupervised ML is further divided into:

1) Clustering
 Raw Data
 Algorithm
 Output
2) Association
 All Values
 Most Frequent Value

Unsupervised Learning Algorithms

 K-means Clustering
 KNN
 Hierarchal Clustering
 Neural Networks\Deep Learning
 Single Value Decomposition
 Distribution Model
 Principal Component Analysis
 Apriori Algorithm

Advantages of Unsupervised ML:

 It is used for more complex task

 Helpful in finding patterns in data
 Saves lot of manual work and expense

Disadvantages of Unsupervised M

 Less Accuracy
 Time Consuming
  More the feature, more the complexity

Feature and Label

Feature: In Machine Learning feature means property of your training data which is used as input.

Label: In Machine Learning label means the output you get from your model after training.

Feature -> Input

Label -> Output

How Scikit-Learn Works:

 Data -> Model -> Prediction

Regression:
A technique for investigating the relationship between independent variables and a dependent variable.
It is used as a method for predictive modelling in machine learning, in which an algorithm is used to
predict continuous outcomes.

Classification:
Classification is a supervised machine learning method where the model tries to predict the correct label
of a given input data. In classification, the model is fully trained using the training data, and then it is
evaluated on test data before being used to perform prediction on new unseen data.

Clustering:
In machine learning too, we often group examples as a first step to understand a subject (data set) in a
machine learning system. Grouping unlabeled examples is called clustering. As the examples are
unlabeled, clustering relies on unsupervised machine learning.

Association:
Rule based machine learning and data mining technique that finds important relations between
variables or features in a data set.

Simple Linear Regression:

 Relationship between two variables
 One independent variable one dependent variable
 Prediction is based on the relationship
 Works in Equation form
 y =a+ bx
- a = constant\intercept
- b = function\slope of x
- x = input
 - y = output
 Y is Dependent Variable
 X is Independent Variable
 Linear = straight line bi-dimentional plot

Multiple Linear Regression

 Relationship between more than two variables
 One dependent variable and more than one independent variable
 Prediction is based on the relationship
 Works in Equation form
 Y = a + b1X1 + b2X2
- a = constant\intercept
- b = function\slope of x
- x = input
- y = output
 Y is Dependent Variable
 X is Independent Variable

Logistics Regression
It is a statistical method that is used for predicting probability of targeted variable. Logistic Regression
makes probability for classification problems that are discrete in nature.

 Logistic Regression tells us the probability

 2 classes based on numeric variable or binary classification
 Classification is based on probability
 If there are two values so, it would be used as classification to classify between the two
values
 If values are more than two so, logistic regression can be used as regression
 No regression line is drawn instead s-shaped function is used
 Works on continuous and discrete data

Linear Regression Logistic Regression

 Line plot  s-shaped function \ s-shaped logistic function

 Use Mean Residual Error  Use Probability “Maximum likelihood”

DECISION TREE CLASSIFIER
Decision tree is a machine learning algorithm based on supervised learning, that can be used for both
regression and classifications problem. The goal is to build up a classifier model that can predict the class
or value of the targeted variable. It can be use as both, classifier and regression.

 We cannot draw lines to separate classes

 Many algorithms are run in Decision Tree
 Many If>Else statements could be run in Decision Tree
 Same as Family Tree
- Until the last value
 Parent Tree\Decision Tree\Decision node\ Root node from where the Tree starts
- Contains conditions
- Split Data
- Child node = Further division of Root node
- Leaf node = Helps us to decide which condition is completed
- Pure node = Last value
 Nested if<Else statement
 Model -> Learn from Data -> Take optional splitting points
- The goal is to take pure leaf node.
- Reduce split
- Try to take maximum gain
 Splitting is done on the basis of information Theory known as Entropy
 Maximum Information gained\don’t care about next splits
 Faster based on selection

Polynomial Regression
Degree of Polynomial:

0 Degree polynomial

Y=constant

1 Degree polynomial

Y=mx+c

2 Degree polynomial

Y=ax2(square)+bx+c

Generalized Polynomial Equation

Y=a0+a1x +

Overfit and Underfit

Model’s performance evaluated based on Accuracy and Generalization.

Accuracy: It states hoe well a model predicts the right variable.

Generalization: It states how well a model behave on a new dataset.

A model is said to be best when it behaves nearly same way on training as well as test data with high
accuracy.

Underfitting means a model has a low accuracy score on training as well as on test data.

It occurs when:

 Due to lack of data

 When we try to build a linear model with non-linear model

Overfitting means a model has a high accuracy score on training as well as on test data.

It occurs when:

 Because of a lot more of data

 When the model fits the data too well

Ideally, when the model makes prediction with zero error, is said to have a good fit on a data.

Confusion Matrix
A confusion matrix is N*N matrix used for evaluating the performance of classification model, where N I
the number of targeted classes. This compares the actual targeted values with predicted targeted value.

Accuracy
The no. of prediction that the model got right.

Accuracy=(TP+TN)/(Tp+TN+FP+FN)

Error Rate
The no. of the prediction that the model got wrong.

Error Rate =(FP+FN)/(TP+TN+FP+FN)

SVM Algorithm
SVM is a machine learning algorithm based on supervised learning, that can be used for both regression
and classification problems..

Types of SVM:

 Linear SVM:
It is used when dataset can be classified into 2 classes using straight line
 Non-Linear SVM
 It is used when dataset cannot be classified into 2 classes using straight line

Kernel Function:

This function takes low dimensional input space and transform it into a higher –dimensional space , it
converts not separable problem into separable problem.

Naïve Bayes Algorithm

Naïve Bayes Classifier is an machine learning algorithm based on supervised learning, that can be used
for solving classification problems.

It is a probabilistic classifier ,which means it predicts on the probability of an object.

Naïve : It’s called naïve because it makes the assumption that all attributes are independent of each
other.

Bayes: It is called bayes because it is dependant on the Principle of Bayes Theorem.

P(A\B)=P(B\A)P(A)

P(B)

Where,

A,B = EVENTS

P(A|B) Probability of A given that B is True

P(B|A) Probability of B given that A is True

P(A)Probability of event A

P(B)Probability of event B

Used in Face Recognition, Weather forecast, News Categorization

Deep Learning
Deep learning is a subset of Machine learning concerned with the algorithms inspired by the structure
and function of human brains…

It is used in Automatic Machine Translation, Colorization of Black and white picture, Fraud news
detection.

Al is a technique that can mimic human behavior.

ML is a technique to achieve AI through algorithms when exposed to more data over time.

DL is a subset of ML in which multilayered neural networks learn from vast amount of data.
 Random Forest Algorithm
Random Forest is a machine learning algorithm based on supervised learning, that can be used for both
regression and classification problems..It is a collection of multiple random decision tree, which is called
Forest.

Structured Query Language

Basics Related to Databases:
 Database is a collection of data in a format that can easily be accessed (Digital).
 A software application used to manage our database is called DBMS (Database Management
System)
 SQL is used to interact with DBMS:
- User  Use SQL To  Interact with DBMS  And DBMS Interact with Database stored in
Cloud Computing Repository
 Types of Databases:
1) Relational Databases 2) Non-Relational Databases
- Data is stored in Tables - Data is stored Unstructured Form
 We use SQL to work with Relational DBMS (RDBMS)

Introduction to SQL(Structured Query Language):

SQL is a programming language used to interact with Relational databases.

It is used to perform CRUD operations:

1) Create
2) Read
3) Update
4) Delete

SQL Data Types:

1. Integer: Used for whole numbers.
2. Binary number: Used to store binary data such as images or files.
3. Char: Fixed-length character string.
4. Datetime: Used to store date and time information.
5. Float: Used for floating-point numbers.
6. Decimal/Numeric: Used for fixed-point numbers with a specified precision and scale.
7. VARCHAR: Variable-length character string.
8. BigInt: Used for large whole numbers.
9. Timestamp: Similar to Datetime, used to store date and time information with higher precision.
10. SMALLINT: Used for small whole numbers.
11. Text/Ntext: Used to store large amounts of text data.
12. Boolean type: Used to store true/false or 1/0 values.
13. Enum: Used to define a set of possible values for a column.
14. Money: Used to store currency values.
15. Nchar/Nvarchar: Unicode character string types, supporting multilingual characters.
16. Real: Used for single-precision floating-point numbers.
17. Precision: Specifies the maximum total number of digits allowed.
18. String: Generic term for character data.
19. Character string types: Include Char, VARCHAR, and others, used to store character data.
20. Data type: Defines the type of data stored in a column.

Signed and Unsigned:

In SQL, "signed" and "unsigned" typically refer to the data type of integer values and how they are
interpreted in terms of their range of values.

1. Signed: Signed integers can represent both positive and negative numbers. The range of values depends
on the number of bits allocated for the integer. For example, a signed integer of 32 bits can represent
values from -2,147,483,648 to 2,147,483,647.
2. Unsigned: Unsigned integers can only represent non-negative numbers (zero or positive). Since they
don't need to store a sign bit, they can represent a wider range of positive values. For example, an
unsigned integer of 32 bits can represent values from 0 to 4,294,967,295.

Types of SQL Commands:

1. Data Definition Language (DDL) Commands:
 CREATE: Used to create database objects like tables, indexes, views, etc.
 ALTER: Modifies the structure of an existing database object.
 DROP: Deletes an existing database object.
 TRUNCATE: Deletes all records from a table, but keeps the table structure intact.
 RENAME: Renames an existing database object.
2. Data Manipulation Language (DML) Commands:
 SELECT: Retrieves data from one or more tables.
 INSERT: Adds new rows of data into a table.
 UPDATE: Modifies existing data in a table.
 DELETE: Removes rows from a table.
3. Data Control Language (DCL) Commands:
 GRANT: Provides specific privileges to database users.
 REVOKE: Removes previously granted privileges from database users.
4. Transaction Control Commands:
 COMMIT: Saves changes made during the current transaction.
 ROLLBACK: Reverts changes made during the current transaction.
 SAVEPOINT: Sets a point within a transaction to which you can later roll back.
5. Data Query Language (DQL) Command:
 SELECT: Retrieves data from one or more tables based on specified criteria.
6. Data Administration Commands:
 ANALYZE: Collects statistics about objects in the database for the query optimizer.
 EXPLAIN: Provides information about how a SQL statement is executed.

Types of SQL Keys:

1. Primary Key:
 A primary key is a column or a set of columns that uniquely identifies each row in a table.
 It must contain unique values and cannot contain NULL values.
 Every table should have a primary key.
 Primary keys are typically used as foreign keys in related tables to establish relationships.
2. Foreign Key:
 A foreign key is a column or a set of columns in one table that refers to the primary key in
another table.
 It enforces referential integrity, ensuring that values in the foreign key column(s) match values
in the primary key column(s) of the referenced table.
 It helps establish relationships between tables.

Types of Constraints:
1. Primary Key Constraint:
 Uniquely identifies each record in a table.
 Ensures values in specified column(s) are unique and not null.
2. Unique Constraint:
 Ensures all values in specified column(s) are unique (null values allowed).
 Multiple unique constraints can be defined in a table.
3. Foreign Key Constraint:
 Establishes a relationship between two tables.
 Ensures referential integrity by enforcing values in one column(s) exist in another table's
primary key or unique key.
4. Check Constraint:
 Enforces domain integrity by limiting values that can be inserted into a column.
 Defines a condition each row must satisfy.
5. Not Null Constraint:
 Ensures a column cannot contain null values.
  Requires every row in the table to have a value for the specified column.

MySql Syntax
 Create table
Create table table_name(
Column1,
Column2
….
);
 Insert
Insert into table_name(column1,column2,…)
Values(value1,value2,…)
 Multiple insert
Insert into table_name(column1,column2,…)
Values
(value1,value2,…),
(value1,value2,…),
(value1,value2,…);
 Constraint
1) Not null
2) Unique
3) Default
4) Check
5) Foreign key
6) Primary key
Create table table name
Specification of value constraint
 Select where
Select *from table name
Select column 1, column2, from table name
 And, Or, Not
Select *from table_ name where and, or, not
Select column 1, column2, from table name where and, or, not
 In
Select column1,column2,….
From table_name
Where column_name IN (value1,value2,…);
 Not in
Select column1,column2,….
From table_name
Where column_name Not IN (value1,value2,…);
 Between
Select column1,column2,….
From table_name
Where column_name Between value1 and value2 And …;
 Not Between
Select column1,column2,….
From table_name
Where column_name not Between value1 and value2 And …;
 Like “s%”
% represent zero, one or multiple character

Select column1,column2,….
From table_name
Where column_name LIKE pattern ;
 Not like
Select column1,column2,….
From table_name
Where column_name Not LIKE pattern ;
 Wildcard Patterns
1) ‘S%’ starts with ‘s’
2) ‘a%’ starts with’ a’
3) ‘%a ‘ end with ‘a’
4) ‘%am%’ have ‘am’ in any position
5) ‘a%m’ starts with ‘a ‘end with ‘m’
6) ‘_a%’ ‘a’ in second position
7) ‘__a%’ ‘ a’ in third position
8) ‘_oy%’ ‘ o’ in second and ‘y’ in third position
9)
 Regular Expression Pattern
1) ‘ ‘
2) ^
3) $
4) […]
5) ^[….]a_z
6) [a-z]
7) P1|p2|p3
select * from Table_name
where column name regexp pattern;
 Order By
1) Select* from Table_name
Order column1,column2,… ASC|DESC;

2) Select* from Table_name

Where column =value

3) Order column1,column2,… ASC|DESC;

select * from practis_two
order by column name;

4) Order column1,column2,… ASC|DESC;

select * from practis_two
order by column1,column2,…;

5) select distinct column1, ,column2

from table_name
order by column_name;

 Is null
Select column1,column2,column3,
From table_name
Where column is null;
 Is not null
Select column1,column2,column3,
From table_name
Where column I nots null;
 Time Series Analysis
Trend:
Trend. The trend shows the general tendency of the data to increase or decrease during a long period of
time. A trend is a smooth, general, long-term, average tendency. It is not always necessary that the
increase or decrease is in the same direction throughout the given period of time.
Trend is based on:
- Pattern of data
- Shows
- General Direction
- Overtime

Stationarity:
Stationarity can be defined in precise mathematical terms, but for our purpose we mean a flat looking
series, without trend, constant variance over time, a constant autocorrelation structure over time and
no periodic fluctuations (seasonality).
- Statistical Property of Time Series which does not change overtime.

Seasonality:
What Is Seasonality? Seasonality is a characteristic of a time series in which the data experiences regular
and predictable changes that recur every calendar year. Any predictable fluctuation or pattern that
recurs or repeats over a one-year period is said to be seasonal.
- Variation in data at specific interval
- It can last Hourly, Daily, Weekly, Monthly, Yearly, Decadely, Century
- Binnning method is best to specify it.
- Clustering algorithms can also be used on seasonality

Cyclic Patterns:
Cyclic. A cycle occurs when the data exhibit rises and falls that are not of a fixed frequency. These
fluctuations are usually due to economic conditions, and are often related to the “business cycle”. The
duration of these fluctuations is usually at least 2 years.
- Fluctuation in data(Time Series)
- Not for a fixed period
- It could not be predicted
Auto-Correlation:
The term autocorrelation refers to the degree of similarity between A) a given time series, and B) a
lagged version of itself, over C) successive time intervals. In other words, autocorrelation is intended to
measure the relationship between a variable's present value and any past values that you may have
access to.
- Measure of Relationship
- Between a Variable’s
- Data’s Current vs Past Times Values

Lag:
The lag time is the time between the two time series you are correlating. If you have time series data at
t=0,1,…,n, then taking the autocorrelation of data sets (0,1),(1,2)… (n−1,n) apart would have a lag time
of 1.
- Time Difference between two related point of times.

White Noise:
White noise is an important concept in time series forecasting. If a time series is white noise, it is a
sequence of random numbers and cannot be predicted. If the series of forecast errors are not white
noise, it suggests improvements could be made to the predictive model.

- Without any pattern

- Random Shuffling

Moving Average:
A moving average is a series of averages, calculated from historic data. Moving averages can be
calculated for any number of time periods, for example a three-month moving average, a seven-day
moving average, or a four-quarter moving average. The basic calculations are the same.
- Smooth out a short term fluctuation
- Long-term pattern will appear
Forecasting:
Time series forecasting is the process of analyzing time series data using statistics and modeling to make
predictions and inform strategic decision-making.
- Prediction based on time

Types of Forecasting Model in Time series Analysis:

- ARIMA autoregressive integrated moving average

- SARIMA Seasonal autoregressive integrated moving average
- ARIMAX autoregressive integrated moving average with exogenous inputs

ARIMA:
An autoregressive integrated moving average, or ARIMA, is a statistical analysis model that uses time
series data to either better understand the data set or to predict future trends.
- Statistical Method for Weather Forecasting
- Used for Prediction
- Works on Autoregressive Integrate
- Autoregressive Integrated Moving Average
- At least 40 data points
- Data should be stationerity

Outliers:
Outliers in time series data are values that significantly differ from the patterns and trends of the other
values in the time series. For example, large numbers of online purchases around holidays or high
numbers of traffic accidents during heavy rainstorms may be detected as outliers in their time series.

Types of Time Series Analysis:

- Descriptive Analysis or EDA of Time Series Data (Different from regular EDA)
- Time Series Forecast for Numeric\Regression (Model are used : ARIMA and SARIMA)
- Time Series Forecast for Categorical\Classification (Model are used : Random Forest,
SVM, Neural Networks and XG Boosting)
- Time Series Decomposition (Big Time Series Data to Small Time Series Data)
 Trend (Long Term Direction of Time Series)
 Seasonality
 Residual (Error of Prediction)
- Spectral Analysis
 Cycles identify cyclic pattern in data

Types of Data about Time and Time series:

- Time Series
- Cross-Sectional Data (one or more variable at a given time point)
- Pooled Data (Combination of Time Series and Cross-Sectional Data)