As the beginning of a journey starts with a single step, this article will aid in the journey of mastering Data Science. With the age of the internet and modern technology, there has been a boom of data and information. This data is a useful and valuable resource in various industries such as healthcare, finance, entertainment and logistics to know about people's patterns and behavior.
Data science is the interpretation of these hoards of data providing insight in making better decisions or making predictive models using machine learning algorithms.
Data Science consists of a five stage life cycle which are:
Capture: This is the collection of raw unstructured data and structured data. This is done through data acquisition, data entry, signal reception and data extraction.
Maintain: Here is where the raw data collected is cleansed and worked on to a provided form of its use. This is attained by data warehousing, data cleansing, data staging, data processing and data architecture.
Process: The prepared data is studied on its ranges, patterns and biases to determine how useful it will be in predictive analysis. This is achieved through data mining, data modeling, clustering/classification and data summarization.
Analyze: The processed data is then examined through various procedures such as exploratory/confirmatory, predictive analysis, regression, text mining and qualitative analysis.
Communicate: The analysis results are plotted into readable formats such as graphs and charts. This is done through data reporting, data visualization, business Intelligence and decision making.
In order to get started in Data Science, some of the prerequisites include:
- Knowledge in Python Language
- Machine Learning
- Statistics
- Knowledge in Databases
- Modelling
Knowledge in Python Language
Python is an interpreted, high-level programming language. One needs to have grasped the basic concepts which enable them to go through the Python libraries necessary for data science exploration and analysis. These are:
Pandas This is used to examine datasets by structuring it into data frames. Also helps in cleaning empty cells, duplicated data and determining relationships in the datasets.
Numpy This is used when working with arrays-preferred as it is faster and consumes less resources compared to Python lists which have a similar function. Handles mathematical manipulation.
Matplotlib This is used to create graphical representations of data and information.
Seaborn This is uses Matplotlib underneath to visualize distributions through graphs.
Machine Learning
This is the use of algorithms to teach machines to function in a certain way as intended and make decisions without human input. These algorithms use data trained by data science. One needs to understand the types of Machine Learning-supervised, unsupervised, semi-supervised, self-supervised and regression.
In addition, understand the concepts of machine learning workflow, evaluation metrics(classification and regression) and the terms-overfitting and underfitting.
Statistics
This is the study of the collection of quantitative data. Some of the basic concepts needed to be familiar with are: mean, median, mode, bias, variance and percentiles. Others include; probability distributions, linear regression, Bayesian statistics, over and under sampling.
Knowledge in Databases
Databases are a collection of structured data and information. One needs to know how to manipulate the data in relational databases using SQL(Structured Query Language). SQL is used in carrying out operations on the database such as querying data, filtering data, grouping data, modifying data and working with table structures.
Modelling
This is the use of the analyzed data to build a machine learning algorithm .The examined data is first split into trained and test data .The trained set is assessed using unknown data points(test data). Based on the evaluation results, the model's parameters can be changed to provide the needed output. This model can then be deployed using Flask(a web application framework) or Amazon Web Services(a cloud computing platform).
I will delve deeper in the coming weeks on Data Science and Machine Learning. Happy learning!