## Introduction:

Data Science is the area of science that extracts meaningful insights from data. Generally, It includes knowledge of programming languages, domain knowledge, mathematical and statistical knowledge. Also, It helps to analyze data and find hidden patterns.

Statistics is a paramount field of study in data science. So, It helps to organize and manage data.

Machine learning trains machines using a model designed using the data available.

Machine learning is of two types:

• Supervised: Supervised learning works on structured data where a targeted variable exists. This further has two techniques:
• Classification: We use this learning program when we want to predict any category.
• Regression: It helps in determining numbers.
• Unsupervised: Unsupervised learning works on unstructured data where no targeted variable exists.

• Visualization and analysis of data.
• Data modeling using complex and efficient algorithms.
• Understanding of data to make a better decision.

## Lifecycle of Data Science:

The main phases of data science are:

• Discovery: First phase of data science lifecycle. So, It involves asking the right questions about the project.
• Data Preparation: Data cleaning, reduction, integration, and transformation are its primary steps.
• Model Planning: Generally, We use different tools to establish relationships between input variables.
• Model Building: In this phase, model building starts using data sets.
• Operationalize: In this phase, we deliver the final report with code and documentation.
• Communication: We will communicate the final result to the team.

## Statistics Concepts Required In Data Science:

Data Scientists have a depth of knowledge of statistics. Basic statistics concepts known to data scientists are:

### Descriptive Statistics:

It helps to analyze raw data. It helps in visualizing data in a meaningful and readable way. Moreover, One uses a plot to visualize data. Inferential statistics helps in finding insights from data analysis.

### Probability:

It helps in determining the likelihood of occurrence of any event. The numerical value of probability lies between 0 and 1. Higher the value, the more likely its occurrence. Further, the probability is of two types:

• Independent Events: Two or more events whose occurrence is independent of each other.
• Conditional Probability: Probability of any event having a relationship with other events.

### Dimensionality Reduction:

We reduce the dimensions of a dataset to make it readable. Dimensionality reduction reduces the complexity of data analysis. So, it offers lesser redundancy, fast computing, and data to store and manage.

### Central Tendency:

It is the value that describes data by central value identification. Generally, Different ways to identify central tendencies are:

• Mean: Average value of data set column.
• Median: Central value of ordered data set.
• Mode: Most repeating value in the data set.
• Skewness: Measures the symmetry of data distribution and determines. It tells whether it has a long tail on either side or not.
• Kurtosis: Defines whether data has normal distribution or has tails.

### Hypothesis Testing:

It tests the result of a survey. There are two types of hypothesis testing:

• Null Hypothesis: It is a general statement and has no relation to surveyed phenomenon.
• Alternate Hypothesis: It is a contradictory statement of the null hypothesis.

### Tests of Significance:

It helps in validating the cited hypothesis. However, Some tests that help in accepting or rejecting the null hypothesis.

• P-value Test
• Z-Test
• T-Test

### Sampling Theory:

Study of relationships that exist between population and sample of it. However, Part of statistics involves data collection, data analysis, and data interpretation. So, We can apply sample theory only to random samples.

### Bayesian Statistics:

A statistical method based on the Bayes Theorem. Moreover, We use it to update the probability for a hypothesis. Generally, It updates the probability based on more evidence and information available.

## Machine Learning Concepts:

• Machine Learning: Subset of Artificial Intelligence. Generally, The machine uses prior knowledge and learns from its previous experiences.
• ML Model: Model built to help machines make predictions based on the dataset.
• Algorithm: Sets of instructions used to build machine learning models.
• Regression: Technique used to determine the relationship between dependent and independent variables. Also, Linear Regression is one of its examples.
• Linear Regression: Most basic linear regression technique. Moreover, It applies to the dataset where there exists a linear relationship between two variables.

• Data Science helps in detecting fraud by using advanced machine learning techniques.
• It helps to avoid any monetary losses.
• Data Science helps in building advanced and intelligent machines.
• It helps in taking quicker and better decisions.
• Data Science helps in analyzing data of the market’s demand and supply. So, It can become a competitive advantage over other companies.

## Challenges Of Data Science:

• Need a huge volume of data for accurate analysis.
• Privacy issues.
• Lacking domain experts.
• Results are not used by businesses and organizations properly.

## Conclusion:

Data Science involves extracting insights from huge amounts of data. Moreover, It involves using various scientific methods, algorithms, and processes. So, Statistics, Deep Learning, Machine Learning are important concepts of data science. However, The biggest challenge that data scientists face is the variety of data. So, Companies like Google use data science and analytics to predict search results. Moreover, Insurance and banking are the other sectors where data science is quite important.

As the data is increasing day by day, data analysts become more important. And thus data science is gaining more popularity. So, Data is the key component of data science. Generally, Dataset is an instance of data helpful in analysis or model building in data science. However, Dataset can be in the form of number, text, video, audio, and image. Generally, Data wrangling is the method of converting data from its raw structure into tidy form. Moreover, Some of these methods are HTML parsing, data structuring, text mining, and more. So, Scaling your features helps in improving the quality and predictability of the model. Moreover, Cross-validation is all about processing machine learning models’ performance for different data sets.